| 1 | /* Copyright (c) 1995, 1997, 1998, 1999, 2000, 2001, 2004, 2005, 2006 |
| 2 | * by Arkkra Enterprises */ |
| 3 | /* All rights reserved */ |
| 4 | /* |
| 5 | * Name: setgrps.c |
| 6 | * |
| 7 | * Description: This file contains functions for setting the relative |
| 8 | * horizontal coordinates of all groups that contain notes |
| 9 | * (grpcont == GC_NOTES) and of all objects in these groups. |
| 10 | * It also sets relative vertical coordinates for the dots |
| 11 | * after notes. |
| 12 | */ |
| 13 | |
| 14 | #include "defines.h" |
| 15 | #include "structs.h" |
| 16 | #include "globals.h" |
| 17 | |
| 18 | struct NOTEPTRS { |
| 19 | struct NOTE *top_p; /* point at a note in top group */ |
| 20 | struct NOTE *bot_p; /* point at same note in bottom group*/ |
| 21 | float wid; /* width of the note head */ |
| 22 | }; |
| 23 | |
| 24 | static struct GRPSYL *procallvoices P((struct MAINLL *mll_p, |
| 25 | struct GRPSYL *gs_p)); |
| 26 | static void proc1or2voices P((struct MAINLL *mll_p, struct STAFF *staff_p, |
| 27 | struct GRPSYL *gs1_p, struct GRPSYL *gs2_p)); |
| 28 | static int compat P((struct NOTEPTRS noteptrs[], struct GRPSYL *gs1_p, |
| 29 | struct GRPSYL *gs2_p)); |
| 30 | static int can_overlap P((struct GRPSYL *gs1_p, struct GRPSYL *gs2_p)); |
| 31 | static void procsome P((struct NOTEPTRS noteptrs[], struct MAINLL *mll_p, |
| 32 | struct STAFF *staff_p, struct GRPSYL *gs1_p, |
| 33 | struct GRPSYL *gs2_p)); |
| 34 | static void procgrace P((struct NOTEPTRS noteptrs[], struct MAINLL *mll_p, |
| 35 | struct STAFF *staff_p, struct GRPSYL *gsnorm_p)); |
| 36 | static void procbunch P((struct NOTEPTRS noteptrs[], struct MAINLL *mll_p, |
| 37 | struct STAFF *staff_p, struct GRPSYL *gs1_p, |
| 38 | struct GRPSYL *gs2_p)); |
| 39 | static void doacc P((struct NOTEPTRS noteptrs[], double halfwide, |
| 40 | double halfhigh, int collinear)); |
| 41 | static int nextacc P((struct NOTEPTRS noteptrs[], int found)); |
| 42 | static void dodot P((struct STAFF *staff_p, struct GRPSYL *gs1_p, |
| 43 | struct GRPSYL *gs2_p, double halfwide, int collinear)); |
| 44 | static void dogrpdot P((struct STAFF *staff_p, struct GRPSYL *gs_p, |
| 45 | struct GRPSYL *ogs_p, double halfwide, int uppermost, |
| 46 | int lowermost, int push)); |
| 47 | static void westwith P((struct GRPSYL *gs_p)); |
| 48 | static void eastwith P((struct GRPSYL *gs_p)); |
| 49 | static void csbstempad P((struct MAINLL *mll_p, struct GRPSYL *gs_p)); |
| 50 | static void proctab P((struct MAINLL *mll_p, struct STAFF *staff_p, |
| 51 | struct GRPSYL *gs1_p)); |
| 52 | static void noterparen P((struct NOTEPTRS noteptrs[], struct GRPSYL *gs1_p, |
| 53 | struct GRPSYL *gs2_p, double halfwide, double halfhigh, |
| 54 | int collinear)); |
| 55 | \f |
| 56 | /* |
| 57 | * Name: setgrps() |
| 58 | * |
| 59 | * Abstract: Find first group on each staff & call procallvoices to process. |
| 60 | * |
| 61 | * Returns: void |
| 62 | * |
| 63 | * Description: This function goes through the chord lists, and for each chord, |
| 64 | * the list of GRPSYLs hanging off it. It finds the first group |
| 65 | * on each staff, and calls procallvoices() to set the relative |
| 66 | * horizontal coordinates of all the note groups on that staff. |
| 67 | */ |
| 68 | |
| 69 | void |
| 70 | setgrps() |
| 71 | |
| 72 | { |
| 73 | struct CHORD *ch_p; /* point at a chord */ |
| 74 | struct GRPSYL *gs1_p; /* point at a group */ |
| 75 | struct MAINLL *mainll_p; /* point at items in main linked list*/ |
| 76 | struct MAINLL *mstaff_p; /* for looking for staff */ |
| 77 | |
| 78 | |
| 79 | debug(16, "setgrps"); |
| 80 | initstructs(); /* clean out old SSV info */ |
| 81 | |
| 82 | /* |
| 83 | * Loop down the main linked list looking for each chord list |
| 84 | * headcell. |
| 85 | */ |
| 86 | for (mainll_p = Mainllhc_p; mainll_p != 0; mainll_p = mainll_p->next) { |
| 87 | |
| 88 | /* keep SSVs up to date */ |
| 89 | if (mainll_p->str == S_SSV) |
| 90 | asgnssv(mainll_p->u.ssv_p); |
| 91 | |
| 92 | if (mainll_p->str != S_CHHEAD) |
| 93 | continue; /* skip everything but chord HC */ |
| 94 | |
| 95 | /* |
| 96 | * Loop through each chord in this list. |
| 97 | */ |
| 98 | for (ch_p = mainll_p->u.chhead_p->ch_p; ch_p != 0; |
| 99 | ch_p = ch_p->ch_p) { |
| 100 | /* |
| 101 | * Loop through the linked list of GRPSYLs hanging off |
| 102 | * this chord. Skip the syllables; just deal with the |
| 103 | * groups. Upon finding the first group on a staff |
| 104 | * (which could be for any of the voices, since not all |
| 105 | * might be present in this chord), call procallvoices |
| 106 | * to process all the note groups. |
| 107 | */ |
| 108 | gs1_p = ch_p->gs_p; |
| 109 | for (;;) { |
| 110 | /* find first group on a staff */ |
| 111 | while (gs1_p != 0 && |
| 112 | gs1_p->grpsyl == GS_SYLLABLE) |
| 113 | gs1_p = gs1_p->gs_p; |
| 114 | if (gs1_p == 0) |
| 115 | break; |
| 116 | |
| 117 | /* find the staff's MLL structure */ |
| 118 | mstaff_p = chmgrp2staffm(mainll_p, gs1_p); |
| 119 | |
| 120 | /* set gs1_p to after this staff's groups */ |
| 121 | gs1_p = procallvoices(mstaff_p, gs1_p); |
| 122 | } |
| 123 | } |
| 124 | } |
| 125 | } |
| 126 | \f |
| 127 | /* |
| 128 | * Name: procallvoices() |
| 129 | * |
| 130 | * Abstract: Process the groups for all the voices on one staff in a chord. |
| 131 | * |
| 132 | * Returns: pointer to the first GRPSYL after these groups, 0 if none |
| 133 | * |
| 134 | * Description: This function is given the GRPSYL for the first (topmost) voice |
| 135 | * that is on this staff in this chord. It finds what other |
| 136 | * GRPSYLs exist. For each of them that is for notes (not rests |
| 137 | * or spaces), it calls proc1or2voices() to process them together |
| 138 | * and/or separately, as needed. This file generally deals only |
| 139 | * with notes, not rests or spaces. But this function also deals |
| 140 | * with rests to the following extent: For both notes and rests, |
| 141 | * there are situations where voice 3 should "stand in" for voice 1 |
| 142 | * or voice 2. This function makes those decisions, and sets pvno. |
| 143 | */ |
| 144 | |
| 145 | static struct GRPSYL * |
| 146 | procallvoices(mll_p, gs_p) |
| 147 | |
| 148 | struct MAINLL *mll_p; /* the MLL item the group is connected to */ |
| 149 | struct GRPSYL *gs_p; /* point at first voice on this staff */ |
| 150 | |
| 151 | { |
| 152 | struct STAFF *staff_p; /* point at staff */ |
| 153 | struct GRPSYL *g_p[MAXVOICES]; /* point at note groups */ |
| 154 | struct GRPSYL *last_p; /* point at last note group */ |
| 155 | struct GRPSYL *g2_p[MAXVOICES]; /* point at note and rest groups */ |
| 156 | struct GRPSYL *gs1_p; /* remember first group */ |
| 157 | struct GRPSYL *gs2_p; /* another GRPSYL pointer */ |
| 158 | int numnonspace; /* number of nonspace GRPSYLs */ |
| 159 | int numgrps; /* how many note groups are here */ |
| 160 | int n; /* loop variable, voices processed */ |
| 161 | |
| 162 | |
| 163 | staff_p = mll_p->u.staff_p; |
| 164 | numgrps = 0; /* no groups found yet */ |
| 165 | last_p = 0; /* no note groups yet */ |
| 166 | gs1_p = gs_p; /* remember first group */ |
| 167 | |
| 168 | /* find all groups in this chord on this staff; remember note groups */ |
| 169 | while (gs_p != 0 && gs_p->staffno == staff_p->staffno && |
| 170 | gs_p->grpsyl == GS_GROUP) { |
| 171 | gs_p->pvno = gs_p->vno; /* init pseudo voice no. to voice no.*/ |
| 172 | if (gs_p->grpcont == GC_NOTES) { |
| 173 | g_p[numgrps++] = gs_p; |
| 174 | last_p = gs_p; |
| 175 | } |
| 176 | gs_p = gs_p->gs_p; |
| 177 | } |
| 178 | |
| 179 | /* |
| 180 | * Before continuing on to process note groups, change voice 3's pvno |
| 181 | * when appropriate. First find all nonspace groups. |
| 182 | */ |
| 183 | numnonspace = 0; /* no nonspace groups found yet */ |
| 184 | gs2_p = gs1_p; |
| 185 | |
| 186 | /* find all nonspace groups in this chord on this staff */ |
| 187 | while (gs2_p != 0 && gs2_p->staffno == staff_p->staffno && |
| 188 | gs2_p->grpsyl == GS_GROUP) { |
| 189 | if (gs2_p->grpcont != GC_SPACE) { |
| 190 | g2_p[numnonspace++] = gs2_p; |
| 191 | } else { |
| 192 | /* |
| 193 | * This is a convenient, though somewhat inappropriate, |
| 194 | * place to process grace groups that precede a space |
| 195 | * group. Ones that precede notes groups will be |
| 196 | * processed in the normal flow, called from procsome. |
| 197 | * They are not allowed before rest groups. |
| 198 | */ |
| 199 | struct NOTEPTRS noteptrs[MAXHAND + 1]; |
| 200 | procgrace(noteptrs, mll_p, staff_p, gs2_p); |
| 201 | } |
| 202 | gs2_p = gs2_p->gs_p; |
| 203 | } |
| 204 | |
| 205 | /* |
| 206 | * If the only nonspace voices are 1 and 3, or 2 and 3, and at least |
| 207 | * one of them is a rest and this is not a tab staff and "ho" was not |
| 208 | * used for either . . . |
| 209 | */ |
| 210 | if (numnonspace == 2 && g2_p[1]->vno == 3 && |
| 211 | (g2_p[0]->grpcont == GC_REST || g2_p[1]->grpcont == GC_REST) && |
| 212 | ! is_tab_staff(staff_p->staffno) && g2_p[0]->ho_usage == HO_NONE && |
| 213 | g2_p[1]->ho_usage == HO_NONE) { |
| 214 | /* |
| 215 | * If v1 is either a rest or stem-up notes and v3 is a rest or |
| 216 | * stem-down notes, let v3 stand in for v2. |
| 217 | */ |
| 218 | if (g2_p[0]->vno == 1 && (g2_p[0]->grpcont == GC_NOTES && |
| 219 | g2_p[0]->stemdir == UP || g2_p[0]->grpcont == GC_REST) && |
| 220 | (g2_p[1]->grpcont == GC_NOTES && g2_p[1]->stemdir == DOWN || |
| 221 | g2_p[1]->grpcont == GC_REST)) { |
| 222 | g2_p[1]->pvno = 2; |
| 223 | } |
| 224 | /* |
| 225 | * If v2 is either a rest or stem-down notes and v3 is a rest or |
| 226 | * stem-up notes, let v3 stand in for v1. |
| 227 | */ |
| 228 | if (g2_p[0]->vno == 2 && (g2_p[0]->grpcont == GC_NOTES && |
| 229 | g2_p[0]->stemdir == DOWN || g2_p[0]->grpcont == GC_REST) && |
| 230 | (g2_p[1]->grpcont == GC_NOTES && g2_p[1]->stemdir == UP || |
| 231 | g2_p[1]->grpcont == GC_REST)) { |
| 232 | g2_p[1]->pvno = 1; |
| 233 | } |
| 234 | } |
| 235 | |
| 236 | /* if there were no note groups on this staff, nothing more to do */ |
| 237 | if (numgrps == 0) |
| 238 | return (gs_p); |
| 239 | |
| 240 | n = 0; /* number of voices processed so far */ |
| 241 | |
| 242 | /* |
| 243 | * If voices 1 and 2 exist and are notes and do not have user specified |
| 244 | * horizontal offsets and this is not a tab staff, handle them together. |
| 245 | * If both voices 1 and 2 have a group here, they will be the first two |
| 246 | * found. Tab staffs should be handled separately because their voices |
| 247 | * never conflict with each other (because of chktabcollision() in |
| 248 | * in setnotes.c). Before checking the offsets, verify that they are |
| 249 | * legal and fix if not. |
| 250 | */ |
| 251 | if (numgrps >= 2 && g_p[0]->vno == 1 && g_p[1]->vno == 2 && |
| 252 | ! is_tab_staff(staff_p->staffno)) { |
| 253 | |
| 254 | vfyoffset(g_p); /* verify and fix */ |
| 255 | |
| 256 | if (g_p[0]->ho_usage == HO_NONE && g_p[1]->ho_usage == HO_NONE){ |
| 257 | proc1or2voices(mll_p, staff_p, g_p[0], g_p[1]); |
| 258 | n = 2; /* processed 2 voices */ |
| 259 | } |
| 260 | } |
| 261 | |
| 262 | /* |
| 263 | * Else, if v1 and v3, or v2 and v3, are notes, and only those two |
| 264 | * exist, and they do not have user specified horizontal offsets and |
| 265 | * this is not a tab staff, and v3's stem dir is compatible, let v3 |
| 266 | * "stand in" for v1 or v2, as the case may be. Handle the two voices |
| 267 | * together. |
| 268 | */ |
| 269 | else if (numgrps == 2 && numnonspace == 2 && |
| 270 | ! is_tab_staff(staff_p->staffno) && g_p[0]->ho_usage == |
| 271 | HO_NONE && g_p[1]->ho_usage == HO_NONE) { |
| 272 | |
| 273 | if (g_p[0]->vno == 1 && g_p[0]->stemdir == UP && |
| 274 | g_p[1]->vno == 3 && g_p[1]->stemdir == DOWN) { |
| 275 | |
| 276 | g_p[1]->pvno = 2; |
| 277 | proc1or2voices(mll_p, staff_p, g_p[0], g_p[1]); |
| 278 | n = 2; /* processed 2 voices */ |
| 279 | |
| 280 | } else if (g_p[0]->vno == 2 && g_p[0]->stemdir == DOWN && |
| 281 | g_p[1]->vno == 3 && g_p[1]->stemdir == UP) { |
| 282 | |
| 283 | g_p[1]->pvno = 1; |
| 284 | proc1or2voices(mll_p, staff_p, g_p[1], g_p[0]); |
| 285 | n = 2; /* processed 2 voices */ |
| 286 | } |
| 287 | } |
| 288 | |
| 289 | /* process any remaining voices individually */ |
| 290 | for ( ; n < numgrps; n++) { |
| 291 | proc1or2voices(mll_p, staff_p, g_p[n], (struct GRPSYL *)0); |
| 292 | } |
| 293 | |
| 294 | /* return the first GRPSYL after the groups we processed */ |
| 295 | return (gs_p); |
| 296 | } |
| 297 | \f |
| 298 | /* |
| 299 | * Name: proc1or2voices() |
| 300 | * |
| 301 | * Abstract: Process a single voice, or voices 1 and 2 together. |
| 302 | * |
| 303 | * Returns: void |
| 304 | * |
| 305 | * Description: This function is given pointers to one or two groups on a |
| 306 | * staff. If it's just one (the second one is a null pointer), |
| 307 | * that group is to be handled alone. If it is two, they are |
| 308 | * voices 1 and 2, since voice 3 is always handled separately. |
| 309 | * (Except that voice 3 can sometimes "stand in" for v1 or v2.) |
| 310 | * In any case, these are always note groups, not rest or space. |
| 311 | * |
| 312 | * The function sets up an array (noteptrs) to point at each |
| 313 | * note in the group(s), figuring out whether the groups overlap |
| 314 | * and, if so, if they are compatible (see below for definition). |
| 315 | * It calls procsome() to set relative horizontal coordinates for |
| 316 | * some notes, which is done either separately for each group or |
| 317 | * both at once, depending on the situation. |
| 318 | */ |
| 319 | |
| 320 | static void |
| 321 | proc1or2voices(mll_p, staff_p, gs1_p, gs2_p) |
| 322 | |
| 323 | struct MAINLL *mll_p; /* the MLL item the group is connected to */ |
| 324 | struct STAFF *staff_p; /* the staff the groups are on */ |
| 325 | register struct GRPSYL *gs1_p, *gs2_p; /* point at groups in this hand */ |
| 326 | |
| 327 | { |
| 328 | /* |
| 329 | * Each structure in this array points at a note. Notes from gs1_p |
| 330 | * are pointed at by top_p, and, when both groups exist, notes |
| 331 | * from gs2_p are pointed at by bot_p. If there's no overlap |
| 332 | * between the groups, there won't be any here either. But if |
| 333 | * the groups "share" notes, the shared notes will be pointed |
| 334 | * at by both. If the groups are "incompatible" (must be |
| 335 | * drawn shifted horizontally to avoid interference), they will |
| 336 | * be done separately and use this array separately, one at a time. |
| 337 | * And in that case, notes from both gs1_p and gs2_p will use top_p, |
| 338 | * in turn. |
| 339 | */ |
| 340 | struct NOTEPTRS noteptrs[MAXHAND + 1]; |
| 341 | |
| 342 | float offset; /* how far to offset incompatible groups */ |
| 343 | int num1; /* number of notes in top group */ |
| 344 | int n; /* loop variable */ |
| 345 | int incompat; /* are groups incompatible (special case) */ |
| 346 | |
| 347 | |
| 348 | /* |
| 349 | * For mrpt, we have nothing to do except set the horizontal group |
| 350 | * coordinates. If the first group is a measure repeat, so is the |
| 351 | * second one, if it exists at all. We set a very small width, as a |
| 352 | * placeholder, because if other staffs have normal notes, we don't |
| 353 | * want the first chord to be abnormally wide because of the mrpt |
| 354 | * symbol. (It will be centered in the measure.) If all the staffs |
| 355 | * have mrpt, abshorz.c will ensure that enough space is left for |
| 356 | * these symbols. |
| 357 | */ |
| 358 | if (is_mrpt(gs1_p)) { |
| 359 | gs1_p->c[RX] = 0; |
| 360 | gs1_p->c[RE] = TEMPMRPTWIDTH / 2.0; |
| 361 | gs1_p->c[RW] = -TEMPMRPTWIDTH / 2.0; |
| 362 | |
| 363 | if (gs2_p != 0) { |
| 364 | gs2_p->c[RX] = 0; |
| 365 | gs2_p->c[RE] = TEMPMRPTWIDTH / 2.0; |
| 366 | gs2_p->c[RW] = -TEMPMRPTWIDTH / 2.0; |
| 367 | } |
| 368 | return; |
| 369 | } |
| 370 | |
| 371 | /* clear out the array */ |
| 372 | for (n = 0; n < NUMELEM(noteptrs); n++) { |
| 373 | noteptrs[n].top_p = 0; |
| 374 | noteptrs[n].bot_p = 0; |
| 375 | noteptrs[n].wid = 0.0; |
| 376 | } |
| 377 | |
| 378 | num1 = gs1_p->nnotes; |
| 379 | |
| 380 | /* set all the "top" group pointers */ |
| 381 | for (n = 0; n < num1; n++) |
| 382 | noteptrs[n].top_p = &gs1_p->notelist[n]; |
| 383 | |
| 384 | /* if there is no "bottom" group, process the first bunch and quit */ |
| 385 | if (gs2_p == 0) { |
| 386 | procsome(noteptrs, mll_p, staff_p, gs1_p, (struct GRPSYL *)0); |
| 387 | |
| 388 | /* if group is rolled, allow room for the roll */ |
| 389 | if (gs1_p->roll != NOITEM) |
| 390 | gs1_p->c[RW] -= ROLLPADDING; |
| 391 | return; |
| 392 | } |
| 393 | |
| 394 | /* |
| 395 | * If the lowest note of the top group is higher than the highest |
| 396 | * note of the bottom group, point at all the bottom notes, |
| 397 | * process both, and quit. Exception: if the inner notes of the |
| 398 | * two groups are on neighboring steps, and the top note of the |
| 399 | * bottom group is on a line and has a dot, and the top group has |
| 400 | * no dots, the groups are to be regarded as if overlapping and |
| 401 | * incompatible. This is because there is no decent way to place |
| 402 | * the dots in this case otherwise. But if, in this neighboring note |
| 403 | * situation, there are no problems with dots, the groups can still be |
| 404 | * handled together here; their stems will be made collinear. When |
| 405 | * the notes are two or more steps apart, there's no problem at all, |
| 406 | * and the groups' X coordinates will line up and equal the chord's. |
| 407 | * Another exception ("else if") is that when the stem of either group |
| 408 | * has been forced the "wrong way" by the user, we require more |
| 409 | * vertical space between the groups. Since we don't know the stem |
| 410 | * lengths yet, we can't do the full job, though. The user may have to |
| 411 | * use "len" or "ho" to avoid a collision. |
| 412 | */ |
| 413 | incompat = NO; |
| 414 | if (noteptrs[num1-1].top_p->stepsup > gs2_p->notelist[0].stepsup) { |
| 415 | if (noteptrs[num1-1].top_p->stepsup == |
| 416 | gs2_p->notelist[0].stepsup + 1 && |
| 417 | gs2_p->notelist[0].stepsup % 2 == 0 && |
| 418 | gs2_p->dots == 0 && |
| 419 | gs1_p->dots > 0) { |
| 420 | incompat = YES; |
| 421 | } else if ((gs1_p->stemdir == DOWN || gs2_p->stemdir == UP) && |
| 422 | noteptrs[num1-1].top_p->stepsup < |
| 423 | gs2_p->notelist[0].stepsup + 3) { |
| 424 | incompat = YES; |
| 425 | } else { |
| 426 | for (n = 0; n < gs2_p->nnotes; n++) |
| 427 | noteptrs[num1+n].bot_p = &gs2_p->notelist[n]; |
| 428 | procsome(noteptrs, mll_p, staff_p, gs1_p, gs2_p); |
| 429 | |
| 430 | /* if a group is rolled, allow room for the roll */ |
| 431 | if (gs1_p->roll != NOITEM) |
| 432 | gs1_p->c[RW] -= ROLLPADDING; |
| 433 | if (gs2_p->roll != NOITEM) |
| 434 | gs2_p->c[RW] -= ROLLPADDING; |
| 435 | return; |
| 436 | } |
| 437 | } |
| 438 | |
| 439 | /* |
| 440 | * There is overlap between the two groups. See if they are |
| 441 | * compatible (also fills in group 2 in noteptrs). If so, |
| 442 | * process the groups together, and return. |
| 443 | */ |
| 444 | if (incompat == NO && compat(noteptrs, gs1_p, gs2_p) == YES) { |
| 445 | procsome(noteptrs, mll_p, staff_p, gs1_p, gs2_p); |
| 446 | |
| 447 | /* if a group is rolled, allow room for the roll */ |
| 448 | if (gs1_p->roll != NOITEM) |
| 449 | gs1_p->c[RW] -= ROLLPADDING; |
| 450 | if (gs2_p->roll != NOITEM) |
| 451 | gs2_p->c[RW] -= ROLLPADDING; |
| 452 | return; |
| 453 | } |
| 454 | |
| 455 | /* |
| 456 | * The fact that we are here means the two groups are not compatible, |
| 457 | * meaning they overlap but can't share note heads. Clear the array |
| 458 | * of any notes from the second group, in case compat() put some there. |
| 459 | */ |
| 460 | for (n = 0; n < NUMELEM(noteptrs); n++) |
| 461 | noteptrs[n].bot_p = 0; |
| 462 | |
| 463 | /* |
| 464 | * It is possible that the groups can at least be given collinear |
| 465 | * stems. For this to be allowed, it must be that the bottom note of |
| 466 | * the top group is on the same step as the top note of the bottom |
| 467 | * group. The top group's note can't have dots, the bottom group's |
| 468 | * can't have accidentals or a roll, and neither can have parentheses, |
| 469 | * because they couldn't be drawn decently. Neither note can have |
| 470 | * another note on a neighboring step. |
| 471 | */ |
| 472 | if (noteptrs[num1-1].top_p->stepsup == gs2_p->notelist[0].stepsup && |
| 473 | |
| 474 | gs1_p->dots == 0 && |
| 475 | |
| 476 | gs2_p->notelist[0].accidental == '\0' && |
| 477 | |
| 478 | gs2_p->roll == NOITEM && |
| 479 | |
| 480 | noteptrs[num1-1].top_p->note_has_paren == NO && |
| 481 | gs2_p->notelist[0].note_has_paren == NO && |
| 482 | |
| 483 | (num1 == 1 || noteptrs[num1-2].top_p->stepsup |
| 484 | > noteptrs[num1-1].top_p->stepsup + 1) && |
| 485 | |
| 486 | (gs2_p->nnotes == 1 || gs2_p->notelist[0].stepsup |
| 487 | > gs2_p->notelist[1].stepsup + 1) ) { |
| 488 | /* |
| 489 | * Since we are not sharing noteheads, the notes of the bottom |
| 490 | * group must be put after the notes of the top group in the |
| 491 | * noteptrs table. Then process them together. |
| 492 | */ |
| 493 | for (n = 0; n < gs2_p->nnotes; n++) |
| 494 | noteptrs[num1+n].bot_p = &gs2_p->notelist[n]; |
| 495 | procsome(noteptrs, mll_p, staff_p, gs1_p, gs2_p); |
| 496 | |
| 497 | /* if top group is rolled, allow room for the roll */ |
| 498 | if (gs1_p->roll != NOITEM) |
| 499 | gs1_p->c[RW] -= ROLLPADDING; |
| 500 | return; |
| 501 | } |
| 502 | |
| 503 | /* |
| 504 | * At this point we know we have to handle the groups separately, and |
| 505 | * then place them. Process the top group now. |
| 506 | */ |
| 507 | procsome(noteptrs, mll_p, staff_p, gs1_p, (struct GRPSYL *)0); |
| 508 | |
| 509 | /* |
| 510 | * Clear the top group out of the array, and fill it with just the |
| 511 | * bottom group, to process them. But mark them as if "top", to |
| 512 | * simplify procsome(). |
| 513 | */ |
| 514 | for (n = 0; n < NUMELEM(noteptrs); n++) |
| 515 | noteptrs[n].top_p = 0; |
| 516 | |
| 517 | /* set all the "top" group pointers even though this is group 2 */ |
| 518 | for (n = 0; n < gs2_p->nnotes; n++) |
| 519 | noteptrs[n].top_p = &gs2_p->notelist[n]; |
| 520 | |
| 521 | procsome(noteptrs, mll_p, staff_p, gs2_p, (struct GRPSYL *)0); |
| 522 | |
| 523 | /* |
| 524 | * Now that we've figured out all the relative horizontal coords for |
| 525 | * the two groups (and everything in them) separately, we need to |
| 526 | * decide how to offset them so they don't overlap. We'll offset |
| 527 | * each the same distance, one right and one left, and apply that |
| 528 | * offset to every horizontal coord of the groups. |
| 529 | */ |
| 530 | /* |
| 531 | * If the groups can be placed so that their rectangles overlap, do it. |
| 532 | * Else if one of the groups is to be rolled and the other is not, the |
| 533 | * one to be rolled must be put on the left. Otherwise, find which |
| 534 | * direction gives minimal offset, but bias the results (0.1) to favor |
| 535 | * putting the top group towards the left, so that the stems will be |
| 536 | * closer to lining up. Set "offset" to the offset to be applied to |
| 537 | * group 1. Group 2's will be -offset. |
| 538 | */ |
| 539 | if (can_overlap(gs1_p, gs2_p) == YES) { |
| 540 | /* top group goes on right; top's offset > 0 */ |
| 541 | if (allsmall(gs1_p, gs1_p) == allsmall(gs2_p, gs2_p)) { |
| 542 | offset = 0.50 * STEPSIZE; |
| 543 | } else { |
| 544 | offset = 0.75 * STEPSIZE; |
| 545 | } |
| 546 | if (gs2_p->roll != NOITEM) |
| 547 | gs2_p->c[RW] -= ROLLPADDING; |
| 548 | } else if (gs1_p->roll != NOITEM && gs2_p->roll == NOITEM) { |
| 549 | /* only top group is rolled; it goes on left; its offset < 0 */ |
| 550 | offset = ( gs2_p->c[RW] - gs1_p->c[RE] ) / 2; |
| 551 | gs1_p->c[RW] -= ROLLPADDING; |
| 552 | } else if (gs1_p->roll == NOITEM && gs2_p->roll != NOITEM) { |
| 553 | /* only bottom is rolled; top goes on right; top's offset > 0 */ |
| 554 | offset = ( gs2_p->c[RE] - gs1_p->c[RW] ) / 2; |
| 555 | gs2_p->c[RW] -= ROLLPADDING; |
| 556 | } else { |
| 557 | /* either both are rolled or neither is; use other criterion */ |
| 558 | if (gs1_p->c[RE] - gs2_p->c[RW] < |
| 559 | gs2_p->c[RE] - gs1_p->c[RW] + 0.1) { |
| 560 | /* top group goes on left; its offset is negative */ |
| 561 | offset = ( gs2_p->c[RW] - gs1_p->c[RE] ) / 2; |
| 562 | if (gs1_p->roll != NOITEM) |
| 563 | gs1_p->c[RW] -= ROLLPADDING; |
| 564 | } else { |
| 565 | /* top group goes on right; its offset is positive */ |
| 566 | offset = ( gs2_p->c[RE] - gs1_p->c[RW] ) / 2; |
| 567 | if (gs2_p->roll != NOITEM) |
| 568 | gs2_p->c[RW] -= ROLLPADDING; |
| 569 | } |
| 570 | } |
| 571 | |
| 572 | /* apply offset to the groups and any preceding grace groups */ |
| 573 | shiftgs(gs1_p, offset); |
| 574 | shiftgs(gs2_p, -offset); |
| 575 | } |
| 576 | \f |
| 577 | /* |
| 578 | * Name: compat() |
| 579 | * |
| 580 | * Abstract: Determine whether two groups in a hand are "compatible". |
| 581 | * |
| 582 | * Returns: YES or NO |
| 583 | * |
| 584 | * Description: This function is given pointers to the two groups in a hand, |
| 585 | * in a situation where they overlap. The noteptrs array has |
| 586 | * just the top group filled in at this point. The function |
| 587 | * figures out whether the two groups are compatible (see block |
| 588 | * comment below), or whether they must be drawn separately and |
| 589 | * offset horizontally. While doing this, it fills in the bottom |
| 590 | * group part of noteptrs. If it returns YES, this has been |
| 591 | * completed. If it returns NO, this may be partially done, |
| 592 | * and the caller should clear out the partially complete bot_p |
| 593 | * part of noteptrs. |
| 594 | */ |
| 595 | |
| 596 | static int |
| 597 | compat(noteptrs, gs1_p, gs2_p) |
| 598 | |
| 599 | struct NOTEPTRS noteptrs[]; /* array of ptrs to notes to process */ |
| 600 | register struct GRPSYL *gs1_p, *gs2_p; /* point at groups in this hand */ |
| 601 | |
| 602 | { |
| 603 | int num1; /* number of notes in top group */ |
| 604 | register int n, k; /* loop variables */ |
| 605 | |
| 606 | |
| 607 | num1 = gs1_p->nnotes; |
| 608 | |
| 609 | /* |
| 610 | * There is overlap between the two groups. Try to match the bottom |
| 611 | * N notes of the top group with the top N notes of the bottom group. |
| 612 | * If all N are "compatible", we can "share" these notes. For two |
| 613 | * groups to be compatible, they must meet the following conditions: |
| 614 | * 1) both basic time values must be half notes, or both must be |
| 615 | * shorter than half notes |
| 616 | * 2) both have no dots or the same number of dots |
| 617 | * 3) the bottom N notes of the top group are the same letters |
| 618 | * and octaves as the top N notes of the bottom group |
| 619 | * 4) no two of these N notes can be on neighboring letters |
| 620 | * 5) for each of the N pairs, the two notes have no accidental |
| 621 | * or the same accidental |
| 622 | * 6) for each of the N pairs, the two notes must have the same |
| 623 | * size and headshape |
| 624 | */ |
| 625 | /* check rule 1 */ |
| 626 | if (gs1_p->basictime < 2 || gs2_p->basictime < 2) |
| 627 | return (NO); |
| 628 | if (gs1_p->basictime == 2 && gs2_p->basictime != 2) |
| 629 | return (NO); |
| 630 | if (gs1_p->basictime != 2 && gs2_p->basictime == 2) |
| 631 | return (NO); |
| 632 | |
| 633 | /* check rule 2 */ |
| 634 | if (gs1_p->dots != gs2_p->dots) |
| 635 | return (NO); |
| 636 | |
| 637 | /* check rules 3, 4, 5, and 6 together */ |
| 638 | /* see if any note in the top group matches the top note in the other*/ |
| 639 | for (n = 0; n < num1; n++) { |
| 640 | if (noteptrs[n].top_p->stepsup == gs2_p->notelist[0].stepsup) |
| 641 | break; |
| 642 | } |
| 643 | if (n == num1) |
| 644 | return (NO); /* didn't find any match */ |
| 645 | |
| 646 | /* starting with this note, verify that it and the rest match */ |
| 647 | for (k = 0; n < num1; k++, n++) { |
| 648 | if (k >= gs2_p->nnotes) /* not enough notes in group 2? */ |
| 649 | return (NO); |
| 650 | if (gs2_p->notelist[k].stepsup != noteptrs[n].top_p->stepsup) |
| 651 | return (NO); |
| 652 | if (k > 0 && |
| 653 | gs2_p->notelist[k-1].stepsup - 1 == gs2_p->notelist[k].stepsup) |
| 654 | return (NO); |
| 655 | if (gs2_p->notelist[k].accidental != noteptrs[n].top_p->accidental) |
| 656 | return (NO); |
| 657 | if (gs2_p->notelist[k].notesize != noteptrs[n].top_p->notesize) |
| 658 | return (NO); |
| 659 | if (gs2_p->notelist[k].headshape != noteptrs[n].top_p->headshape) |
| 660 | return (NO); |
| 661 | |
| 662 | /* this note matches; set up noteptrs */ |
| 663 | noteptrs[n].bot_p = &gs2_p->notelist[k]; |
| 664 | } |
| 665 | |
| 666 | /* |
| 667 | * The fact that we made it to here means all the overlapping notes |
| 668 | * matched. So fill the rest of group 2's note pointers. |
| 669 | */ |
| 670 | for ( ; k < gs2_p->nnotes; k++, n++) |
| 671 | noteptrs[n].bot_p = &gs2_p->notelist[k]; |
| 672 | /* |
| 673 | * It is possible that, although the overlapping notes' headshapes |
| 674 | * match, some of the characters are mirrors of each other due to the |
| 675 | * opposite stem dir. In these cases, group 2 rules. So overwrite the |
| 676 | * notes in group 1. If the lowest note in group 1 has to be changed, |
| 677 | * that could affect the RS of group 1, so change that too. |
| 678 | * Also, while doing this, if any of these notes or their accs have |
| 679 | * parens in one group but not the other, erase those parens. |
| 680 | */ |
| 681 | n -= k; |
| 682 | for (k = 0; n < num1; k++, n++) { |
| 683 | gs1_p->notelist[n].headchar = gs2_p->notelist[k].headchar; |
| 684 | gs1_p->notelist[n].headfont = gs2_p->notelist[k].headfont; |
| 685 | gs1_p->notelist[n].c[RN] = gs2_p->notelist[k].c[RN]; |
| 686 | gs1_p->notelist[n].c[RS] = gs2_p->notelist[k].c[RS]; |
| 687 | |
| 688 | if (gs1_p->notelist[n].note_has_paren != |
| 689 | gs2_p->notelist[k].note_has_paren) { |
| 690 | gs1_p->notelist[n].note_has_paren = NO; |
| 691 | gs2_p->notelist[k].note_has_paren = NO; |
| 692 | } |
| 693 | if (gs1_p->notelist[n].acc_has_paren != |
| 694 | gs2_p->notelist[k].acc_has_paren) { |
| 695 | gs1_p->notelist[n].acc_has_paren = NO; |
| 696 | gs2_p->notelist[k].acc_has_paren = NO; |
| 697 | } |
| 698 | } |
| 699 | gs1_p->c[RS] = gs2_p->notelist[k - 1].c[RS]; |
| 700 | |
| 701 | return (YES); |
| 702 | } |
| 703 | \f |
| 704 | /* |
| 705 | * Name: can_overlap() |
| 706 | * |
| 707 | * Abstract: Decides whether incompatible groups' rectangles can overlap. |
| 708 | * |
| 709 | * Returns: YES or NO |
| 710 | * |
| 711 | * Description: This function is given two incompatible groups in a hand. It |
| 712 | * decides whether they can be placed such that their rectangles |
| 713 | * overlap. This arrangement is where the first group is to the |
| 714 | * right of the second group, and the stems are about 3 stepsizes |
| 715 | * apart. The noteheads must be separated enough vertically so |
| 716 | * that they don't collide, and various other things must also be |
| 717 | * true for this to work. |
| 718 | */ |
| 719 | |
| 720 | static int |
| 721 | can_overlap(gs1_p, gs2_p) |
| 722 | |
| 723 | struct GRPSYL *gs1_p, *gs2_p; /* point at group(s) in this hand */ |
| 724 | |
| 725 | { |
| 726 | int notedist; /* steps between two notes (absolute value) */ |
| 727 | int n, k; /* loop counters */ |
| 728 | |
| 729 | |
| 730 | /* |
| 731 | * First, ensure that no note heads would collide. We don't yet know |
| 732 | * whether any will be on the "wrong" side of their stem. This is not |
| 733 | * too common and would rarely help things, so for now we assume the |
| 734 | * worst case, which is that all are on the "correct" side and thus |
| 735 | * have the potential of colliding with the other group's notes. |
| 736 | */ |
| 737 | for (n = 0; n < gs1_p->nnotes; n++) { |
| 738 | for (k = 0; k < gs2_p->nnotes; k++) { |
| 739 | notedist = abs(gs1_p->notelist[n].stepsup - |
| 740 | gs2_p->notelist[k].stepsup); |
| 741 | |
| 742 | /* never allow closer than 2 steps */ |
| 743 | if (notedist < 2) |
| 744 | return (NO); |
| 745 | |
| 746 | /* if either is double whole, don't allow less than 3 */ |
| 747 | if ((gs1_p->basictime == 0 || gs2_p->basictime == 0) && |
| 748 | notedist < 3) |
| 749 | return (NO); |
| 750 | } |
| 751 | } |
| 752 | |
| 753 | /* neither group can have slashes */ |
| 754 | if (gs1_p->slash_alt > 0 || gs2_p->slash_alt > 0) |
| 755 | return (NO); |
| 756 | |
| 757 | /* the first group can't have accidentals */ |
| 758 | for (n = 0; n < gs1_p->nnotes; n++) { |
| 759 | if (gs1_p->notelist[n].accidental != '\0') |
| 760 | return (NO); |
| 761 | } |
| 762 | |
| 763 | /* the first group can't any preceding grace groups */ |
| 764 | if (gs1_p->prev != 0 && gs1_p->prev->grpvalue == GV_ZERO) |
| 765 | return (NO); |
| 766 | |
| 767 | /* the first group can't have a roll unless the second group has one */ |
| 768 | if (gs1_p->roll != NOITEM && gs2_p->roll == NOITEM) |
| 769 | return (NO); |
| 770 | |
| 771 | /* the second group can't have any dots */ |
| 772 | if (gs2_p->dots > 0) |
| 773 | return (NO); |
| 774 | |
| 775 | /* the second group can't have any flags */ |
| 776 | if (gs2_p->basictime >= 8 && gs2_p->beamloc == NOITEM) |
| 777 | return (NO); |
| 778 | |
| 779 | /* neither group can have a stem forced the "wrong" way */ |
| 780 | if (gs1_p->stemdir == DOWN || gs2_p->stemdir == UP) |
| 781 | return (NO); |
| 782 | |
| 783 | /* |
| 784 | * At this point we know we can overlap. |
| 785 | */ |
| 786 | return (YES); |
| 787 | } |
| 788 | \f |
| 789 | /* |
| 790 | * Name: procsome() |
| 791 | * |
| 792 | * Abstract: Sets coords for group(s) and their associated grace groups. |
| 793 | * |
| 794 | * Returns: void |
| 795 | * |
| 796 | * Description: This function calls procbunch() to set the horizontal coords |
| 797 | * for the given group(s) and their notes, etc. Then it calls |
| 798 | * procgrace() to deal with any grace groups preceding these |
| 799 | * group(s) and adjust the main group(s)' west coordinates to. |
| 800 | * contain the grace groups. |
| 801 | */ |
| 802 | |
| 803 | static void |
| 804 | procsome(noteptrs, mll_p, staff_p, gs1_p, gs2_p) |
| 805 | |
| 806 | struct NOTEPTRS noteptrs[]; /* array of ptrs to notes to process */ |
| 807 | struct MAINLL *mll_p; /* the MLL item the group is connected to */ |
| 808 | struct STAFF *staff_p; /* the staff the groups are connected to */ |
| 809 | struct GRPSYL *gs1_p, *gs2_p; /* point at group(s) in this hand */ |
| 810 | |
| 811 | { |
| 812 | /* process the normal group(s) */ |
| 813 | procbunch(noteptrs, mll_p, staff_p, gs1_p, gs2_p); |
| 814 | |
| 815 | /* process any grace groups preceding first normal group */ |
| 816 | procgrace(noteptrs, mll_p, staff_p, gs1_p); |
| 817 | |
| 818 | /* process any grace groups preceding second normal group, if exists */ |
| 819 | if (gs2_p != 0) |
| 820 | procgrace(noteptrs, mll_p, staff_p, gs2_p); |
| 821 | } |
| 822 | \f |
| 823 | /* |
| 824 | * Name: procgrace() |
| 825 | * |
| 826 | * Abstract: Sets coords for grace groups and adjusts normal group's west. |
| 827 | * |
| 828 | * Returns: void |
| 829 | * |
| 830 | * Description: This function loops leftward from the given normal group, |
| 831 | * calling procbunch() for each grace group, and adjusting the |
| 832 | * normal group's west coordinate accordingly. |
| 833 | */ |
| 834 | |
| 835 | static void |
| 836 | procgrace(noteptrs, mll_p, staff_p, gsnorm_p) |
| 837 | |
| 838 | struct NOTEPTRS noteptrs[]; /* array of ptrs to notes to process */ |
| 839 | struct MAINLL *mll_p; /* the MLL item the group is connected to */ |
| 840 | struct STAFF *staff_p; /* the staff the groups are connected to */ |
| 841 | struct GRPSYL *gsnorm_p; /* point at the normal group to start from */ |
| 842 | |
| 843 | { |
| 844 | struct GRPSYL *gs_p; /* point at a grace group */ |
| 845 | struct GRPSYL *right_p; /* point at the group to the right of this */ |
| 846 | int n; /* loop variable */ |
| 847 | |
| 848 | |
| 849 | /* |
| 850 | * Loop through any grace groups preceding the normal group, working |
| 851 | * right to left. Call procbunch() for each. Upon return, set |
| 852 | * the grace group's x,e,w relative to the normal group's x, and |
| 853 | * alter the west coordinate of the normal group to include them. |
| 854 | */ |
| 855 | right_p = gsnorm_p; |
| 856 | for (gs_p = gsnorm_p->prev; gs_p != 0 && gs_p->grpvalue == GV_ZERO; |
| 857 | gs_p = gs_p->prev) { |
| 858 | /* clear noteptrs, and resetup for this grace group */ |
| 859 | /* note: grace groups are always notes, not rests or spaces */ |
| 860 | for (n = 0; n < MAXHAND + 1; n++) { |
| 861 | noteptrs[n].top_p = 0; |
| 862 | noteptrs[n].bot_p = 0; |
| 863 | } |
| 864 | /* set all the "top" group pointers */ |
| 865 | for (n = 0; n < gs_p->nnotes; n++) |
| 866 | noteptrs[n].top_p = &gs_p->notelist[n]; |
| 867 | |
| 868 | procbunch(noteptrs, mll_p, staff_p, gs_p, (struct GRPSYL *)0); |
| 869 | |
| 870 | gs_p->c[RX] = right_p->c[RW] - gs_p->c[RE]; |
| 871 | gs_p->c[RW] += gs_p->c[RX]; |
| 872 | gs_p->c[RE] += gs_p->c[RX]; |
| 873 | |
| 874 | gsnorm_p->c[RW] = gs_p->c[RW]; |
| 875 | right_p = gs_p; |
| 876 | } |
| 877 | } |
| 878 | \f |
| 879 | /* |
| 880 | * Name: procbunch() |
| 881 | * |
| 882 | * Abstract: Sets relative horizontal coords of note heads, accs, & dots. |
| 883 | * |
| 884 | * Returns: void |
| 885 | * |
| 886 | * Description: This function figures out which note heads in the given |
| 887 | * group(s) need to be put on the "wrong" side of the stem to |
| 888 | * avoid overlapping. Then it sets all note heads' horizontal |
| 889 | * coords. It calls doacc() to find and store the positions |
| 890 | * for the accidentals, dodot() for the dots. It sets RW and |
| 891 | * RE for the group(s), also taking flags into consideration. |
| 892 | */ |
| 893 | |
| 894 | /* |
| 895 | * This macro checks the n'th structure in noteptrs. If the top group has |
| 896 | * a note there, it returns a pointer to that note, else it returns the |
| 897 | * bottom pointer, which may or may not be 0. |
| 898 | */ |
| 899 | #define GETPTR(n) (noteptrs[n].top_p != 0 ? \ |
| 900 | noteptrs[n].top_p : noteptrs[n].bot_p) |
| 901 | |
| 902 | static void |
| 903 | procbunch(noteptrs, mll_p, staff_p, gs1_p, gs2_p) |
| 904 | |
| 905 | struct NOTEPTRS noteptrs[]; /* array of ptrs to notes to process */ |
| 906 | struct MAINLL *mll_p; /* the MLL item the group is connected to */ |
| 907 | struct STAFF *staff_p; /* the staff the groups are connected to */ |
| 908 | struct GRPSYL *gs1_p, *gs2_p; /* point at group(s) in this hand */ |
| 909 | |
| 910 | { |
| 911 | int normhead[MAXHAND + 1]; /* position of note heads */ |
| 912 | float gwide; /* width of any note in these groups */ |
| 913 | float nwide; /* width of a particular note */ |
| 914 | float maxwide; /* max of gwide for the two groups */ |
| 915 | float ghigh; /* height of any note in these groups*/ |
| 916 | float nhigh; /* height of a particular note */ |
| 917 | float g1wide, g2wide; /* gwide for the two groups */ |
| 918 | float maxhigh; /* max of ghigh for the two groups */ |
| 919 | float flagwidth; /* width of a flag */ |
| 920 | float rh; /* relative horizontal of a note */ |
| 921 | int collinear; /* are the 2 groups' stems collinear? */ |
| 922 | register int k, n; /* loop variables */ |
| 923 | int size; |
| 924 | |
| 925 | |
| 926 | /* |
| 927 | * If this is a tablature staff, call a special function to handle it, |
| 928 | * and return. Voices on tab staffs are handled one at a time, so |
| 929 | * gs2_p will never be used for them. |
| 930 | */ |
| 931 | if (is_tab_staff(staff_p->staffno)) { |
| 932 | proctab(mll_p, staff_p, gs1_p); |
| 933 | return; |
| 934 | } |
| 935 | |
| 936 | collinear = NO; /* assume not collinear stems */ |
| 937 | |
| 938 | /* |
| 939 | * "Normal" position of a note head means to the left of the stem |
| 940 | * for an upward stem, and right for downward. When two notes in a |
| 941 | * group are on neighboring letters, one of the note heads has to be |
| 942 | * in "abnormal" position so that they don't collide. Shared |
| 943 | * note heads must always be in normal position. (The fact |
| 944 | * that no two of them can be on neighboring letters is enforced |
| 945 | * when checking for compatibility of groups.) |
| 946 | */ |
| 947 | /* |
| 948 | * See if there are any shared notes first. |
| 949 | */ |
| 950 | for (n = 0; noteptrs[n].top_p != 0; n++) { |
| 951 | if (noteptrs[n].bot_p != 0) |
| 952 | break; /* found a shared note */ |
| 953 | } |
| 954 | |
| 955 | if (noteptrs[n].top_p != 0) { |
| 956 | /* |
| 957 | * There are shared notes, and n indexes to the first one |
| 958 | * (starting from the top). Set this first one to normal. |
| 959 | * First work upwards from there, reversing normality |
| 960 | * whenever there are neighboring notes, setting back to |
| 961 | * normal otherwise. Then work downwards from there, doing |
| 962 | * the same. |
| 963 | */ |
| 964 | normhead[n] = YES; |
| 965 | for (k = n - 1 ; k >= 0; k--) { |
| 966 | if (noteptrs[k+1].top_p->stepsup == |
| 967 | noteptrs[ k ].top_p->stepsup - 1) |
| 968 | normhead[k] = ! normhead[k+1]; |
| 969 | else |
| 970 | normhead[k] = YES; |
| 971 | } |
| 972 | for (k = n + 1 ; noteptrs[k].bot_p != 0; k++) { |
| 973 | if (noteptrs[k-1].bot_p->stepsup == |
| 974 | noteptrs[ k ].bot_p->stepsup + 1) |
| 975 | normhead[k] = ! normhead[k-1]; |
| 976 | else |
| 977 | normhead[k] = YES; |
| 978 | } |
| 979 | } else { |
| 980 | /* |
| 981 | * There are no shared notes. It may even be that there's only |
| 982 | * one group. In each group, the note that's opposite the stem |
| 983 | * must be normal, and then we go down the list of other notes |
| 984 | * in the group, reversing normality whenever there are |
| 985 | * neighboring notes, and setting back to normal otherwise. |
| 986 | * There's a special concern if the bottom note of the top |
| 987 | * group is on the neighboring letter to the top note of the |
| 988 | * bottom group, or if it is on the same letter. In that case, |
| 989 | * we want to offset the groups slightly, such that their stems |
| 990 | * are collinear, so set that flag. |
| 991 | */ |
| 992 | /* the first group's stem could go either way */ |
| 993 | if (gs1_p->stemdir == UP) { |
| 994 | normhead[n-1] = YES; /* bottom note normal */ |
| 995 | for (k = n - 2; k >= 0; k--) { |
| 996 | if (noteptrs[k+1].top_p->stepsup == |
| 997 | noteptrs[ k ].top_p->stepsup - 1) |
| 998 | normhead[k] = ! normhead[k+1]; |
| 999 | else |
| 1000 | normhead[k] = YES; |
| 1001 | } |
| 1002 | } else { /* stemdir == DOWN */ |
| 1003 | normhead[0] = YES; /* top note normal */ |
| 1004 | for (k = 1; k < n; k++) { |
| 1005 | if (noteptrs[k-1].top_p->stepsup == |
| 1006 | noteptrs[ k ].top_p->stepsup + 1) |
| 1007 | normhead[k] = ! normhead[k-1]; |
| 1008 | else |
| 1009 | normhead[k] = YES; |
| 1010 | } |
| 1011 | } |
| 1012 | |
| 1013 | /* the second group's stem (if it exists) must go down */ |
| 1014 | if (gs2_p != 0) { |
| 1015 | normhead[n] = YES; /* top note normal */ |
| 1016 | for (k = n + 1; noteptrs[k].bot_p != 0; k++) { |
| 1017 | if (noteptrs[k-1].bot_p->stepsup == |
| 1018 | noteptrs[ k ].bot_p->stepsup + 1) |
| 1019 | normhead[k] = ! normhead[k-1]; |
| 1020 | else |
| 1021 | normhead[k] = YES; |
| 1022 | } |
| 1023 | |
| 1024 | collinear = (noteptrs[n-1].top_p->stepsup <= |
| 1025 | noteptrs[ n ].bot_p->stepsup + 1); |
| 1026 | } |
| 1027 | } |
| 1028 | |
| 1029 | /* |
| 1030 | * Set gwide and ghigh to be the biggest values of any note in the top |
| 1031 | * group, also storing the width of each note for later use. |
| 1032 | */ |
| 1033 | gwide = ghigh = 0.0; |
| 1034 | for (n = 0; noteptrs[n].top_p != 0; n++) { |
| 1035 | size = noteptrs[n].top_p->notesize == GS_NORMAL ? |
| 1036 | DFLT_SIZE : SMALLSIZE; |
| 1037 | nwide = width(noteptrs[n].top_p->headfont, size, |
| 1038 | noteptrs[n].top_p->headchar); |
| 1039 | noteptrs[n].wid = nwide; |
| 1040 | if (nwide > gwide) { |
| 1041 | gwide = nwide; |
| 1042 | } |
| 1043 | nhigh = height(noteptrs[n].top_p->headfont, size, |
| 1044 | noteptrs[n].top_p->headchar); |
| 1045 | if (nhigh > ghigh) { |
| 1046 | ghigh = nhigh; |
| 1047 | } |
| 1048 | } |
| 1049 | |
| 1050 | /* remember these values, for comparing to the other group (if any) */ |
| 1051 | maxwide = g1wide = gwide; /* widest group so far */ |
| 1052 | maxhigh = ghigh; /* highest group so far */ |
| 1053 | |
| 1054 | if (gs1_p->basictime <= 1) { |
| 1055 | gs1_p->stemx = 0.0; /* center the imaginary stem */ |
| 1056 | } else { |
| 1057 | gs1_p->stemx = gs1_p->stemdir == UP ? gwide / 2 : -gwide / 2; |
| 1058 | } |
| 1059 | |
| 1060 | for (n = 0; noteptrs[n].top_p != 0; n++) { |
| 1061 | nwide = noteptrs[n].wid; |
| 1062 | |
| 1063 | if (normhead[n] == YES) { |
| 1064 | /* |
| 1065 | * The note head is in normal position, so usually its |
| 1066 | * relative x coord is 0, and west and east are half a |
| 1067 | * width off. But if the note is smaller than the |
| 1068 | * group's max, and there is a stem, and the note is |
| 1069 | * not shared by the other group, the note needs to |
| 1070 | * be off center so that it touches the stem. |
| 1071 | */ |
| 1072 | if (nwide != gwide && gs1_p->basictime >= 2 && |
| 1073 | noteptrs[n].bot_p == 0) { |
| 1074 | if (gs1_p->stemdir == UP) { |
| 1075 | noteptrs[n].top_p->c[RE] = gwide / 2; |
| 1076 | noteptrs[n].top_p->c[RX] = |
| 1077 | gwide / 2 - nwide / 2; |
| 1078 | noteptrs[n].top_p->c[RW] = |
| 1079 | gwide / 2 - nwide; |
| 1080 | } else { /* DOWN */ |
| 1081 | noteptrs[n].top_p->c[RW] = -gwide / 2; |
| 1082 | noteptrs[n].top_p->c[RX] = |
| 1083 | -gwide / 2 + nwide / 2; |
| 1084 | noteptrs[n].top_p->c[RE] = |
| 1085 | -gwide / 2 + nwide; |
| 1086 | } |
| 1087 | } else { |
| 1088 | noteptrs[n].top_p->c[RX] = 0; |
| 1089 | noteptrs[n].top_p->c[RW] = -nwide / 2; |
| 1090 | noteptrs[n].top_p->c[RE] = nwide / 2; |
| 1091 | } |
| 1092 | } else { |
| 1093 | /* |
| 1094 | * The note head is in abnormal position. Its relative |
| 1095 | * x coord, and west and east, depend on which way the |
| 1096 | * stem is going. Smaller than normal notes need to |
| 1097 | * be placed differently regardless of whether stemed. |
| 1098 | * In all case, adjust by W_NORMAL*POINT, the width of |
| 1099 | * the stem, so that the note overlays the stem. |
| 1100 | */ |
| 1101 | if (nwide != gwide) { |
| 1102 | if (gs1_p->stemdir == UP) { |
| 1103 | noteptrs[n].top_p->c[RW] = |
| 1104 | gwide / 2 - W_NORMAL * POINT; |
| 1105 | noteptrs[n].top_p->c[RX] = |
| 1106 | gwide / 2 + nwide / 2 |
| 1107 | - W_NORMAL * POINT; |
| 1108 | noteptrs[n].top_p->c[RE] = |
| 1109 | gwide / 2 + nwide |
| 1110 | - W_NORMAL * POINT; |
| 1111 | } else { /* DOWN */ |
| 1112 | noteptrs[n].top_p->c[RE] = |
| 1113 | W_NORMAL * POINT - gwide / 2; |
| 1114 | noteptrs[n].top_p->c[RX] = |
| 1115 | W_NORMAL * POINT |
| 1116 | - gwide / 2 - nwide /2; |
| 1117 | noteptrs[n].top_p->c[RW] = |
| 1118 | W_NORMAL * POINT |
| 1119 | - gwide / 2 - nwide; |
| 1120 | } |
| 1121 | } else { |
| 1122 | if (gs1_p->stemdir == UP) { |
| 1123 | noteptrs[n].top_p->c[RX] = |
| 1124 | nwide - W_NORMAL * POINT; |
| 1125 | noteptrs[n].top_p->c[RW] = |
| 1126 | nwide * 0.5 - W_NORMAL * POINT; |
| 1127 | noteptrs[n].top_p->c[RE] = |
| 1128 | nwide * 1.5 - W_NORMAL * POINT; |
| 1129 | } else { /* DOWN */ |
| 1130 | noteptrs[n].top_p->c[RX] = |
| 1131 | W_NORMAL * POINT - nwide; |
| 1132 | noteptrs[n].top_p->c[RW] = |
| 1133 | W_NORMAL * POINT - nwide * 1.5; |
| 1134 | noteptrs[n].top_p->c[RE] = |
| 1135 | W_NORMAL * POINT - nwide * 0.5; |
| 1136 | } |
| 1137 | } |
| 1138 | } |
| 1139 | } |
| 1140 | |
| 1141 | /* |
| 1142 | * If there is a bottom group, get note head character width for |
| 1143 | * it, find where in noteptrs that group starts, then loop through |
| 1144 | * it, setting coords. While doing this, set the group's |
| 1145 | * horizontal coords. |
| 1146 | */ |
| 1147 | g2wide = 0.0; /* to avoid useless 'used before set' warning */ |
| 1148 | if (gs2_p != 0) { |
| 1149 | /* skip by notes that are only in the top group */ |
| 1150 | for (n = 0; noteptrs[n].bot_p == 0; n++) |
| 1151 | ; |
| 1152 | /* |
| 1153 | * Set gwide and ghigh to be the biggest values of any note in |
| 1154 | * the bottom group, also storing the width of each note for |
| 1155 | * later use. If the note is shared between groups, the width |
| 1156 | * has already been stored in noteptrs[].wid, so we don't have |
| 1157 | * to recalculate it. |
| 1158 | */ |
| 1159 | gwide = ghigh = 0.0; |
| 1160 | for ( ; noteptrs[n].bot_p != 0; n++) { |
| 1161 | size = noteptrs[n].bot_p->notesize == GS_NORMAL ? |
| 1162 | DFLT_SIZE : SMALLSIZE; |
| 1163 | if (noteptrs[n].wid == 0.0) { |
| 1164 | nwide = width(noteptrs[n].bot_p->headfont, size, |
| 1165 | noteptrs[n].bot_p->headchar); |
| 1166 | noteptrs[n].wid = nwide; |
| 1167 | } else { |
| 1168 | nwide = noteptrs[n].wid; |
| 1169 | } |
| 1170 | if (nwide > gwide) { |
| 1171 | gwide = nwide; |
| 1172 | } |
| 1173 | nhigh = height(noteptrs[n].bot_p->headfont, size, |
| 1174 | noteptrs[n].bot_p->headchar); |
| 1175 | if (nhigh > ghigh) { |
| 1176 | ghigh = nhigh; |
| 1177 | } |
| 1178 | } |
| 1179 | g2wide = gwide; |
| 1180 | if (gs2_p->basictime <= 1) { |
| 1181 | gs2_p->stemx = 0.0; /* center the imaginary stem */ |
| 1182 | } else { |
| 1183 | gs2_p->stemx = gs2_p->stemdir == UP ? gwide / 2 |
| 1184 | : -gwide / 2; |
| 1185 | } |
| 1186 | |
| 1187 | /* if groups have different note head sizes, adjust maxes */ |
| 1188 | if (gwide > maxwide) |
| 1189 | maxwide = gwide; |
| 1190 | if (ghigh > maxhigh) |
| 1191 | maxhigh = ghigh; |
| 1192 | |
| 1193 | for (n = 0; noteptrs[n].bot_p == 0; n++) |
| 1194 | ; |
| 1195 | for ( ; noteptrs[n].bot_p != 0; n++) { |
| 1196 | nwide = noteptrs[n].wid; |
| 1197 | |
| 1198 | if (normhead[n] == YES) { |
| 1199 | /* |
| 1200 | * The note head is in normal position, so its |
| 1201 | * relative x coord is 0, and west and east are |
| 1202 | * half a width off. But if the note is smaller |
| 1203 | * than the widest note in the group and there |
| 1204 | * is a stem, and the note is not shared by the |
| 1205 | * other group, the note needs to be off center |
| 1206 | * so that it touches the stem. |
| 1207 | */ |
| 1208 | if (nwide != gwide && gs2_p->basictime >= 2 && |
| 1209 | noteptrs[n].top_p == 0) { |
| 1210 | noteptrs[n].bot_p->c[RW] = -gwide / 2; |
| 1211 | noteptrs[n].bot_p->c[RX] = |
| 1212 | -gwide / 2 + nwide / 2; |
| 1213 | noteptrs[n].bot_p->c[RE] = |
| 1214 | -gwide / 2 + nwide; |
| 1215 | } else { |
| 1216 | noteptrs[n].bot_p->c[RX] = 0; |
| 1217 | noteptrs[n].bot_p->c[RW] = -nwide * 0.5; |
| 1218 | noteptrs[n].bot_p->c[RE] = nwide * 0.5; |
| 1219 | } |
| 1220 | } else { |
| 1221 | /* |
| 1222 | * The note head is in abnormal position. Its |
| 1223 | * relative x coord, and west and east, depend |
| 1224 | * on which way the stem is going, but the |
| 1225 | * stem must always be down in group 2. Smaller |
| 1226 | * than normal notes need to be placed |
| 1227 | * differently regardless of whether stemed. |
| 1228 | */ |
| 1229 | if (nwide != gwide) { |
| 1230 | noteptrs[n].bot_p->c[RE] = |
| 1231 | W_NORMAL * POINT - gwide / 2; |
| 1232 | noteptrs[n].bot_p->c[RX] = |
| 1233 | W_NORMAL * POINT |
| 1234 | - gwide / 2 - nwide /2; |
| 1235 | noteptrs[n].bot_p->c[RW] = |
| 1236 | W_NORMAL * POINT |
| 1237 | - gwide / 2 - nwide; |
| 1238 | } else { |
| 1239 | noteptrs[n].bot_p->c[RX] = |
| 1240 | W_NORMAL * POINT - nwide; |
| 1241 | noteptrs[n].bot_p->c[RW] = |
| 1242 | W_NORMAL * POINT - nwide * 1.5; |
| 1243 | noteptrs[n].bot_p->c[RE] = |
| 1244 | W_NORMAL * POINT - nwide * 0.5; |
| 1245 | } |
| 1246 | } |
| 1247 | } |
| 1248 | } |
| 1249 | |
| 1250 | /* find position of accidentals */ |
| 1251 | doacc(noteptrs, maxwide / 2, maxhigh / 2, collinear); |
| 1252 | |
| 1253 | /* find position of dots after notes */ |
| 1254 | dodot(staff_p, gs1_p, gs2_p, maxwide / 2, collinear); |
| 1255 | |
| 1256 | /* find position of right parentheses around notes */ |
| 1257 | noterparen(noteptrs, gs1_p, gs2_p, maxwide/2, maxhigh/2, collinear); |
| 1258 | |
| 1259 | /* |
| 1260 | * Set RX for the group(s) to 0 for now if stems are offset (the |
| 1261 | * normal case), or to the appropriate value if stems are collinear. |
| 1262 | * If we only have one group it will thus be set to 0 now, though |
| 1263 | * later, if there's an incompatible group next to it, this coord |
| 1264 | * and all others will be adjusted. |
| 1265 | */ |
| 1266 | if (collinear) { |
| 1267 | gs1_p->c[RX] = (W_NORMAL * POINT - maxwide) / 2; |
| 1268 | gs2_p->c[RX] = (maxwide - W_NORMAL * POINT) / 2; |
| 1269 | } else { |
| 1270 | gs1_p->c[RX] = 0; |
| 1271 | if (gs2_p != 0) |
| 1272 | gs2_p->c[RX] = 0; |
| 1273 | } |
| 1274 | |
| 1275 | /* |
| 1276 | * Set the western boundaries for the group(s). |
| 1277 | */ |
| 1278 | /* |
| 1279 | * Init the group's RW to 0. Then loop through the notes, finding the |
| 1280 | * westernmost thing associated with a note, and leaving the group's RW |
| 1281 | * set to that. |
| 1282 | */ |
| 1283 | gs1_p->c[RW] = 0; |
| 1284 | for (k = 0; k < gs1_p->nnotes; k++) { |
| 1285 | rh = notehorz(gs1_p, &gs1_p->notelist[k], RW); |
| 1286 | if (rh < gs1_p->c[RW]) |
| 1287 | gs1_p->c[RW] = rh; |
| 1288 | } |
| 1289 | /* |
| 1290 | * If the stem is down on a half note or shorter that is to have |
| 1291 | * slashes through its stem, make sure there is room for the slashes. |
| 1292 | */ |
| 1293 | if (gs1_p->slash_alt > 0 && gs1_p->stemdir == DOWN && |
| 1294 | gs1_p->basictime >= 2) { |
| 1295 | gwide = g1wide; |
| 1296 | /* if position of stem minus slash room < current west . . . */ |
| 1297 | if (-gwide / 2 - SLASHPAD < gs1_p->c[RW]) |
| 1298 | gs1_p->c[RW] = -gwide / 2 - SLASHPAD; |
| 1299 | } |
| 1300 | westwith(gs1_p); /* expand RW for "with" list if needbe*/ |
| 1301 | gs1_p->c[RW] -= gs1_p->padding; /* add user requested padding */ |
| 1302 | |
| 1303 | /* add the pad parameter that user wants for this voice */ |
| 1304 | gs1_p->c[RW] -= vvpath(gs1_p->staffno, gs1_p->vno, PAD)->pad; |
| 1305 | |
| 1306 | csbstempad(mll_p, gs1_p); /* cross staff beaming may need space */ |
| 1307 | gs1_p->c[RW] += gs1_p->c[RX]; /* shift by RX, in case RX isn't 0 */ |
| 1308 | |
| 1309 | /* |
| 1310 | * If group 2 exists, do the same for it. However, in the slash |
| 1311 | * section, we know the stem must be down, so no need to check that. |
| 1312 | */ |
| 1313 | if (gs2_p != 0) { |
| 1314 | gs2_p->c[RW] = 0; |
| 1315 | for (k = 0; k < gs2_p->nnotes; k++) { |
| 1316 | rh = notehorz(gs2_p, &gs2_p->notelist[k], RW); |
| 1317 | if (rh < gs2_p->c[RW]) |
| 1318 | gs2_p->c[RW] = rh; |
| 1319 | } |
| 1320 | if (gs2_p->slash_alt > 0 && gs2_p->basictime >= 2) { |
| 1321 | gwide = g2wide; |
| 1322 | /* if pos of stem minus slash room < current west . .*/ |
| 1323 | if (-gwide / 2 - SLASHPAD < gs2_p->c[RW]) |
| 1324 | gs2_p->c[RW] = -gwide / 2 - SLASHPAD; |
| 1325 | } |
| 1326 | westwith(gs2_p); |
| 1327 | gs2_p->c[RW] -= gs2_p->padding; |
| 1328 | gs2_p->c[RW] -= vvpath(gs2_p->staffno, gs2_p->vno, PAD)->pad; |
| 1329 | csbstempad(mll_p, gs2_p); |
| 1330 | gs2_p->c[RW] += gs2_p->c[RX]; |
| 1331 | } |
| 1332 | |
| 1333 | /* |
| 1334 | * Set the eastern boundaries for the group(s). |
| 1335 | */ |
| 1336 | /* |
| 1337 | * Init the group's RE to 0. Then loop through the notes, finding the |
| 1338 | * easternmost thing associated with a note, and leaving the group's RE |
| 1339 | * set to that. |
| 1340 | */ |
| 1341 | gs1_p->c[RE] = 0; |
| 1342 | for (k = 0; k < gs1_p->nnotes; k++) { |
| 1343 | rh = notehorz(gs1_p, &gs1_p->notelist[k], RE); |
| 1344 | if (rh > gs1_p->c[RE]) |
| 1345 | gs1_p->c[RE] = rh; |
| 1346 | } |
| 1347 | /* |
| 1348 | * Add in any padding needed for ties, slurs, and bends. Also add room |
| 1349 | * for alternations if there are any. |
| 1350 | */ |
| 1351 | gs1_p->c[RE] += tieslurpad(staff_p, gs1_p); |
| 1352 | if (gs1_p->slash_alt < 0 && gs1_p->beamloc == STARTITEM) |
| 1353 | gs1_p->c[RE] += ALTPAD; |
| 1354 | /* |
| 1355 | * If the stem is up and a flag is needed, and the east boundary |
| 1356 | * doesn't yet contain it, adjust the east boundary so the flag will |
| 1357 | * fit. |
| 1358 | */ |
| 1359 | if (gs1_p->stemdir == UP && gs1_p->basictime >= 8 && |
| 1360 | gs1_p->beamloc == NOITEM) { |
| 1361 | flagwidth = width(FONT_MUSIC, gs1_p->grpsize == GS_NORMAL ? |
| 1362 | DFLT_SIZE : SMALLSIZE, C_UPFLAG); |
| 1363 | if (gs1_p->notelist[0].c[RE] + flagwidth > gs1_p->c[RE]) |
| 1364 | gs1_p->c[RE] = gs1_p->notelist[0].c[RE] + flagwidth; |
| 1365 | } |
| 1366 | /* |
| 1367 | * If the stem is up on a half note or shorter that is to have slashes |
| 1368 | * through its stem, make sure there's room for the slashes. |
| 1369 | */ |
| 1370 | if (gs1_p->slash_alt > 0 && gs1_p->stemdir == UP && |
| 1371 | gs1_p->basictime >= 2) { |
| 1372 | gwide = g1wide; |
| 1373 | /* if position of stem plus slash room > current east . . . */ |
| 1374 | if (gwide / 2 + SLASHPAD > gs1_p->c[RE]) |
| 1375 | gs1_p->c[RE] = gwide / 2 + SLASHPAD; |
| 1376 | } |
| 1377 | /* |
| 1378 | * Expand RE some more if need be to accommodate the "with" list. Then |
| 1379 | * shift it over by RX, in case RX isn't 0. |
| 1380 | */ |
| 1381 | eastwith(gs1_p); |
| 1382 | gs1_p->c[RE] += gs1_p->c[RX]; |
| 1383 | |
| 1384 | /* |
| 1385 | * If group 2 exists, do the same for it. However, the stem is always |
| 1386 | * down, so any flags will always already fit. For the same reason, |
| 1387 | * slashes don't need to be considered. |
| 1388 | */ |
| 1389 | if (gs2_p != 0) { |
| 1390 | gs2_p->c[RE] = 0; |
| 1391 | for (k = 0; k < gs2_p->nnotes; k++) { |
| 1392 | rh = notehorz(gs2_p, &gs2_p->notelist[k], RE); |
| 1393 | if (rh > gs2_p->c[RE]) |
| 1394 | gs2_p->c[RE] = rh; |
| 1395 | } |
| 1396 | gs2_p->c[RE] += tieslurpad(staff_p, gs2_p); |
| 1397 | if (gs2_p->slash_alt < 0 && gs2_p->beamloc == STARTITEM) |
| 1398 | gs2_p->c[RE] += ALTPAD; |
| 1399 | eastwith(gs2_p); |
| 1400 | gs2_p->c[RE] += gs2_p->c[RX]; |
| 1401 | } |
| 1402 | } |
| 1403 | \f |
| 1404 | /* |
| 1405 | * Name: doacc() |
| 1406 | * |
| 1407 | * Abstract: Finds horizontal position for each accidental in group(s). |
| 1408 | * |
| 1409 | * Returns: void |
| 1410 | * |
| 1411 | * Description: This function loops through all the accidentals belonging |
| 1412 | * to notes in the group(s) it is given. It figures out where |
| 1413 | * to place them horizontally to avoid overlap, and stores the |
| 1414 | * relative west coord of each in NOTE.waccr. For each group, |
| 1415 | * it uses the appropriate size of accidentals (based on normal |
| 1416 | * versus cue/grace), and places them appropriately, considering |
| 1417 | * also the size of the notes. However, if there are two groups, |
| 1418 | * the note head sizes could be different. The halfwide and |
| 1419 | * halfhigh passed in are supposed to be the right size for the |
| 1420 | * bigger of the two sizes, and accidentals will not be packed |
| 1421 | * as tightly against the other notes. This doesn't hurt, and |
| 1422 | * isn't worth the trouble to do it "right". |
| 1423 | * |
| 1424 | * This function takes into account parentheses around accidentals. |
| 1425 | * Its algorithm treats them as part of the accidental. Also, when |
| 1426 | * there are parentheses around the note, it handles the left |
| 1427 | * parentheses the same way: if there is also an accidental, it |
| 1428 | * treats it as part of it; otherwise the paren is handled like an |
| 1429 | * accidental itself. |
| 1430 | */ |
| 1431 | |
| 1432 | /* this fudge factor prevents roundoff error from causing overlap */ |
| 1433 | #define FUDGE (.01) |
| 1434 | |
| 1435 | /* when CSS applies to a note or acc, move it by this much */ |
| 1436 | #define CSS_OFF (CSS_STEPS * STEPSIZE) |
| 1437 | |
| 1438 | static void |
| 1439 | doacc(noteptrs, halfwide, halfhigh, collinear) |
| 1440 | |
| 1441 | struct NOTEPTRS noteptrs[]; /* array of ptrs to notes to process */ |
| 1442 | double halfwide; /* half of max of width & height of (notes */ |
| 1443 | double halfhigh; /* in group 1, notes in group 2) */ |
| 1444 | int collinear; /* are stems collinear? */ |
| 1445 | |
| 1446 | { |
| 1447 | /* |
| 1448 | * Each structure in this table represents either a note head that |
| 1449 | * is farther left than normal, or an accidental. A note head |
| 1450 | * could be too far left for one of two reasons: either it was |
| 1451 | * forced to be on the left ("wrong") side of a stem that points |
| 1452 | * down, or it is a normal note in the top group when the stems are |
| 1453 | * collinear. In the collinear case, to make this function easier, |
| 1454 | * we start out regarding the bottom group as being normal, and |
| 1455 | * the top group as being shifted left one note head, and we figure |
| 1456 | * everything relative to the bottom group. But at the end we adjust |
| 1457 | * waccr so that every accidental is relative to its own group, like |
| 1458 | * it's supposed to be. |
| 1459 | * |
| 1460 | * The coordinates define the rectangle that surrounds the note or acc, |
| 1461 | * including standard padding, even on note heads, which don't |
| 1462 | * normally have padding. First the notes are put into this table; |
| 1463 | * then the accidentals, one at a time, making sure they don't |
| 1464 | * overlap things already in the table. |
| 1465 | * To see if the accidental being added overlaps, first its north |
| 1466 | * and south are tested. All previous rectangles that are "out of |
| 1467 | * its way" vertically are marked not "relevant"; the others are |
| 1468 | * marked "relevant". As positions are tried, right to left, positions |
| 1469 | * that fail to avoid overlap are marked "tried". |
| 1470 | * |
| 1471 | * After the correct position is found for an accidental, there is a |
| 1472 | * special case for flats and double flats to take advantage of their |
| 1473 | * shape and let them pack tighter. |
| 1474 | */ |
| 1475 | struct { |
| 1476 | float n, s, e, w; /* boundaries of a rectangle */ |
| 1477 | short relevant; /* is rectangle relevant? */ |
| 1478 | short tried; /* have we tried this one yet? */ |
| 1479 | } rectab[2 * MAXHAND + 1]; /* enough for all notes & accidentals*/ |
| 1480 | |
| 1481 | struct NOTE *note_p; /* point at a note */ |
| 1482 | int reclim; /* index after last rectangle in tab */ |
| 1483 | float north, south, east, west; /* relative coords of new accidental */ |
| 1484 | float accasc, accdesc; /* ascent & descent of accidental */ |
| 1485 | float accwidth; /* width of new accidental */ |
| 1486 | float parenwidth; /* width of note's left parenthesis */ |
| 1487 | float parenv; /* half the vertical size of paren */ |
| 1488 | float totwidth; /* width of acc plus paren */ |
| 1489 | int overlap; /* does our acc overlap existing ones*/ |
| 1490 | int try; /* which element of rectab to try */ |
| 1491 | int found; /* accs/parens found so far */ |
| 1492 | int k, j; /* loop variables */ |
| 1493 | int size; |
| 1494 | float horfn, verfn; /* horz & vert flat/nat notch sizes */ |
| 1495 | float savehorfn; /* save original horfn */ |
| 1496 | |
| 1497 | |
| 1498 | reclim = 0; /* table initially empty */ |
| 1499 | |
| 1500 | /* |
| 1501 | * Loop through noteptrs, finding all notes that are left of normal |
| 1502 | * position, entering them in rectab. Include padding around them. |
| 1503 | * First loop through all notes, finding ones that are on the left |
| 1504 | * side of a down stem; then, if stems are collinear, loop through |
| 1505 | * the top group, finding all normal notes. |
| 1506 | */ |
| 1507 | for (k = 0; (note_p = GETPTR(k)) != 0; k++) { |
| 1508 | if (note_p->c[RX] < 0) { |
| 1509 | rectab[reclim].n = note_p->c[RY] + halfhigh + STDPAD; |
| 1510 | rectab[reclim].s = note_p->c[RY] - halfhigh - STDPAD; |
| 1511 | rectab[reclim].e = note_p->c[RE] + STDPAD; |
| 1512 | rectab[reclim].w = note_p->c[RW] - STDPAD; |
| 1513 | if (note_p->stepsup >= CSS_STEPS / 2) { |
| 1514 | rectab[reclim].n += CSS_OFF; |
| 1515 | rectab[reclim].s += CSS_OFF; |
| 1516 | } else if (note_p->stepsup <= -CSS_STEPS / 2) { |
| 1517 | rectab[reclim].n -= CSS_OFF; |
| 1518 | rectab[reclim].s -= CSS_OFF; |
| 1519 | } |
| 1520 | reclim++; |
| 1521 | } |
| 1522 | } |
| 1523 | if (collinear) { |
| 1524 | for (k = 0; (note_p = noteptrs[k].top_p) != 0; k++) { |
| 1525 | if (note_p->c[RX] == 0) { |
| 1526 | rectab[reclim].n = note_p->c[RY] + halfhigh |
| 1527 | + STDPAD; |
| 1528 | rectab[reclim].s = note_p->c[RY] - halfhigh |
| 1529 | - STDPAD; |
| 1530 | rectab[reclim].e = W_NORMAL * POINT |
| 1531 | - halfwide + STDPAD; |
| 1532 | rectab[reclim].w = W_NORMAL * POINT |
| 1533 | - 3 * halfwide - STDPAD; |
| 1534 | if (note_p->stepsup >= CSS_STEPS / 2) { |
| 1535 | rectab[reclim].n += CSS_OFF; |
| 1536 | rectab[reclim].s += CSS_OFF; |
| 1537 | } else if (note_p->stepsup <= -CSS_STEPS / 2) { |
| 1538 | rectab[reclim].n -= CSS_OFF; |
| 1539 | rectab[reclim].s -= CSS_OFF; |
| 1540 | } |
| 1541 | reclim++; |
| 1542 | } |
| 1543 | } |
| 1544 | } |
| 1545 | |
| 1546 | /* prevent false "may be used before set" lint warning */ |
| 1547 | verfn = savehorfn = 0.0; |
| 1548 | |
| 1549 | /* |
| 1550 | * Loop through all notes, find the ones with accs or parens. Find |
| 1551 | * where the accs and parens will fit, storing that info in waccr, and |
| 1552 | * adding them to rectab. Call a function so that we loop in the |
| 1553 | * proper order. |
| 1554 | */ |
| 1555 | for (found = 0, k = nextacc(noteptrs, found); k != -1; |
| 1556 | found++, k = nextacc(noteptrs, found)) { |
| 1557 | note_p = GETPTR(k); |
| 1558 | /* get dimensions of accidental if there is one */ |
| 1559 | if (note_p->accidental != '\0') { |
| 1560 | accdimen(note_p, &accasc, &accdesc, &accwidth); |
| 1561 | } else { |
| 1562 | accwidth = accasc = accdesc = 0.0; |
| 1563 | } |
| 1564 | /* get dimensions of note's left paren, if there is one */ |
| 1565 | if (note_p->note_has_paren == YES) { |
| 1566 | size = (note_p->notesize == GS_NORMAL ? |
| 1567 | DFLT_SIZE : SMALLSIZE); |
| 1568 | parenwidth = width(FONT_TR, size, '('); |
| 1569 | parenv = height(FONT_TR, size, '(') / 2.0; |
| 1570 | } else { |
| 1571 | parenwidth = parenv = 0.0; |
| 1572 | } |
| 1573 | /* set the north, south, and width of what we have found */ |
| 1574 | north = note_p->c[RY] + MAX(accasc, parenv); |
| 1575 | south = note_p->c[RY] - MAX(accdesc, parenv); |
| 1576 | if (note_p->stepsup >= CSS_STEPS / 2) { |
| 1577 | north += CSS_OFF; |
| 1578 | south += CSS_OFF; |
| 1579 | } else if (note_p->stepsup <= -CSS_STEPS / 2) { |
| 1580 | north -= CSS_OFF; |
| 1581 | south -= CSS_OFF; |
| 1582 | } |
| 1583 | totwidth = accwidth + parenwidth; |
| 1584 | |
| 1585 | /* |
| 1586 | * For each rectangle in rectab, decide whether (based on |
| 1587 | * its vertical coords) it could possibly overlap with our |
| 1588 | * new accidental. If it's totally above or below ours, it |
| 1589 | * can't. We allow a slight overlap (FUDGE) so that round |
| 1590 | * off errors don't stop us from packing things as tightly |
| 1591 | * as possible. |
| 1592 | */ |
| 1593 | for (j = 0; j < reclim; j++) { |
| 1594 | if (rectab[j].s + FUDGE > north || |
| 1595 | rectab[j].n < south + FUDGE) |
| 1596 | rectab[j].relevant = NO; |
| 1597 | else |
| 1598 | rectab[j].relevant = YES; |
| 1599 | } |
| 1600 | |
| 1601 | /* |
| 1602 | * Mark that none of the relevant rectangles' boundaries have |
| 1603 | * been tried yet for positioning our acc. |
| 1604 | */ |
| 1605 | for (j = 0; j < reclim; j++) { |
| 1606 | if (rectab[j].relevant == YES) |
| 1607 | rectab[j].tried = NO; |
| 1608 | } |
| 1609 | |
| 1610 | /* |
| 1611 | * Set up first trial position for this acc., just to the |
| 1612 | * left of normal notes, allowing padding. |
| 1613 | */ |
| 1614 | east = - halfwide - STDPAD; |
| 1615 | west = east - totwidth; |
| 1616 | |
| 1617 | /* |
| 1618 | * Keep trying positions for this acc, working right to |
| 1619 | * left. When we find one that doesn't overlap an existing |
| 1620 | * rectangle, break. This has to succeed at some point, |
| 1621 | * at the leftmost rectangle position if not earlier. |
| 1622 | */ |
| 1623 | for (;;) { |
| 1624 | overlap = NO; |
| 1625 | for (j = 0; j < reclim; j++) { |
| 1626 | /* ignore ones too far north or south */ |
| 1627 | if (rectab[j].relevant == NO) |
| 1628 | continue; |
| 1629 | |
| 1630 | /* if all west or east, okay; else overlap */ |
| 1631 | if (rectab[j].w + FUDGE <= east && |
| 1632 | rectab[j].e >= west + FUDGE) { |
| 1633 | overlap = YES; |
| 1634 | break; |
| 1635 | } |
| 1636 | } |
| 1637 | |
| 1638 | /* if no rectangle overlapped, we found a valid place*/ |
| 1639 | if (overlap == NO) |
| 1640 | break; |
| 1641 | |
| 1642 | /* |
| 1643 | * Something overlapped, so we have to try again. |
| 1644 | * Find the eastermost relevant west rectangle boundary |
| 1645 | * that hasn't been tried already, to use as the next |
| 1646 | * trial position for our acc's east. |
| 1647 | */ |
| 1648 | try = -1; |
| 1649 | for (j = 0; j < reclim; j++) { |
| 1650 | /* ignore ones too far north or south */ |
| 1651 | if (rectab[j].relevant == NO || |
| 1652 | rectab[j].tried == YES) |
| 1653 | continue; |
| 1654 | |
| 1655 | /* |
| 1656 | * If this is the first relevant one we haven't |
| 1657 | * tried, or if this is farther east than the |
| 1658 | * easternmost so far, save it as being the |
| 1659 | * new easternmost so far. |
| 1660 | */ |
| 1661 | if (try == -1 || rectab[j].w > rectab[try].w) |
| 1662 | try = j; |
| 1663 | } |
| 1664 | |
| 1665 | if (try == -1) |
| 1666 | pfatal("bug in doacc()"); |
| 1667 | |
| 1668 | /* |
| 1669 | * Mark this one as having been tried (for next time |
| 1670 | * around, if necessary). Set new trial values for |
| 1671 | * east and west of our acc. |
| 1672 | */ |
| 1673 | rectab[try].tried = YES; |
| 1674 | east = rectab[try].w; |
| 1675 | west = east - totwidth; |
| 1676 | |
| 1677 | } /* end of while loop trying positions for this acc */ |
| 1678 | |
| 1679 | /* |
| 1680 | * We found the correct position for the new acc. However, for |
| 1681 | * flats, double flats & nats, we would like a notch to be taken |
| 1682 | * out of the upper right corner of their rectangle, in effect, |
| 1683 | * since there's nothing there but white space. This can only |
| 1684 | * be done if the acc is not already right next to the group. |
| 1685 | */ |
| 1686 | if (note_p->accidental == '&' || note_p->accidental == 'B' || |
| 1687 | note_p->accidental == 'n') { |
| 1688 | /* get notch size; if paren, add width to horz */ |
| 1689 | if (note_p->accidental == 'n') { |
| 1690 | horfn = 1.4 * STEPSIZE; /* horizontal notch */ |
| 1691 | verfn = 1.6 * STEPSIZE; /* vertical notch */ |
| 1692 | } else { |
| 1693 | horfn = 1.5 * STEPSIZE; /* horizontal notch */ |
| 1694 | verfn = 2.8 * STEPSIZE; /* vertical notch */ |
| 1695 | } |
| 1696 | if (note_p->notesize == GS_SMALL) { |
| 1697 | horfn *= SM_FACTOR; |
| 1698 | verfn *= SM_FACTOR; |
| 1699 | } |
| 1700 | if (note_p->acc_has_paren) { |
| 1701 | size = (note_p->notesize == GS_NORMAL ? |
| 1702 | DFLT_SIZE : SMALLSIZE); |
| 1703 | horfn += width(FONT_TR, size, ')'); |
| 1704 | } |
| 1705 | savehorfn = horfn; /* may need it later */ |
| 1706 | /* |
| 1707 | * If notch width is bigger than the max possible dist |
| 1708 | * we could move the acc (we would overwrite the note), |
| 1709 | * reduce it to be the space available. |
| 1710 | */ |
| 1711 | if (horfn > - east - halfwide - STDPAD) |
| 1712 | horfn = - east - halfwide - STDPAD; |
| 1713 | |
| 1714 | /* only attempt the shift if > 0 width available */ |
| 1715 | if (horfn > 0.0) { |
| 1716 | /* |
| 1717 | * The useable notch size is horfn by verfn. |
| 1718 | * We'd like to move the acc to the right by |
| 1719 | * horfn. We can only do this if the space is |
| 1720 | * unoccupied that is immediately to the right |
| 1721 | * of the acc, of width = horfn and height = |
| 1722 | * (height of acc) - verfn. (If only part of |
| 1723 | * that space is available, we won't bother |
| 1724 | * trying to use it.) So check whether any |
| 1725 | * existing rectangle overlaps that space. |
| 1726 | */ |
| 1727 | overlap = NO; |
| 1728 | for (j = 0; j < reclim; j++) { |
| 1729 | if (rectab[j].s + FUDGE <= north - verfn && |
| 1730 | rectab[j].n - FUDGE >= south && |
| 1731 | rectab[j].w + FUDGE <= east + horfn && |
| 1732 | rectab[j].e - FUDGE >= east) { |
| 1733 | overlap = YES; |
| 1734 | break; |
| 1735 | } |
| 1736 | } |
| 1737 | /* |
| 1738 | * If the space is free, move the acc to the |
| 1739 | * right by HORFN. |
| 1740 | */ |
| 1741 | if (overlap == NO) { |
| 1742 | west += horfn; |
| 1743 | east += horfn; |
| 1744 | } else { |
| 1745 | /* |
| 1746 | * All right, let's try again with 1/2 |
| 1747 | * of the previous horfn. |
| 1748 | */ |
| 1749 | horfn /= 2.0; |
| 1750 | overlap = NO; |
| 1751 | for (j = 0; j < reclim; j++) { |
| 1752 | if (rectab[j].s + FUDGE <= north - verfn && |
| 1753 | rectab[j].n - FUDGE >= south && |
| 1754 | rectab[j].w + FUDGE <= east + horfn && |
| 1755 | rectab[j].e - FUDGE >= east) { |
| 1756 | overlap = YES; |
| 1757 | break; |
| 1758 | } |
| 1759 | } |
| 1760 | if (overlap == NO) { |
| 1761 | west += horfn; |
| 1762 | east += horfn; |
| 1763 | } |
| 1764 | } |
| 1765 | } |
| 1766 | } |
| 1767 | |
| 1768 | /* |
| 1769 | * We have the final position for the new acc. Enter it into |
| 1770 | * rectab. But for naturals, we don't want to reserve the |
| 1771 | * lower left corner, where there is nothing but white space; |
| 1772 | * so in that case, put two overlapping entries in rectab to |
| 1773 | * account for the rest of the space. Naturals are symmetrical, |
| 1774 | * so we can use the same horfn and verfn as were calculated |
| 1775 | * above for the upper right corner. |
| 1776 | */ |
| 1777 | if (note_p->accidental == 'n') { |
| 1778 | /* upper part of natural */ |
| 1779 | rectab[reclim].n = north; |
| 1780 | rectab[reclim].s = south + verfn; |
| 1781 | rectab[reclim].e = east; |
| 1782 | rectab[reclim].w = west; |
| 1783 | reclim++; |
| 1784 | |
| 1785 | /* right hand part of natural */ |
| 1786 | rectab[reclim].n = north; |
| 1787 | rectab[reclim].s = south; |
| 1788 | rectab[reclim].e = east; |
| 1789 | rectab[reclim].w = west + savehorfn; |
| 1790 | } else { |
| 1791 | /* some other accidental; reserve the whole rectangle*/ |
| 1792 | rectab[reclim].n = north; |
| 1793 | rectab[reclim].s = south; |
| 1794 | rectab[reclim].e = east; |
| 1795 | rectab[reclim].w = west; |
| 1796 | } |
| 1797 | reclim++; |
| 1798 | |
| 1799 | /* |
| 1800 | * Store the acc's west in waccr in the NOTE structure for |
| 1801 | * whichever groups have this note. Store wlparen when there |
| 1802 | * is a left paren on the note. |
| 1803 | */ |
| 1804 | if (noteptrs[k].top_p != 0) { |
| 1805 | if (note_p->note_has_paren == YES) |
| 1806 | noteptrs[k].top_p->wlparen = west; |
| 1807 | if (note_p->accidental != '\0') |
| 1808 | noteptrs[k].top_p->waccr = west + parenwidth; |
| 1809 | } |
| 1810 | if (noteptrs[k].bot_p != 0) { |
| 1811 | if (note_p->note_has_paren == YES) |
| 1812 | noteptrs[k].bot_p->wlparen = west; |
| 1813 | if (note_p->accidental != '\0') |
| 1814 | noteptrs[k].bot_p->waccr = west + parenwidth; |
| 1815 | } |
| 1816 | |
| 1817 | } /* end of loop for each accidental */ |
| 1818 | |
| 1819 | /* |
| 1820 | * Finally, if the stems were collinear, we have to adjust waccr for |
| 1821 | * all the notes of the top group, so that it's relative to the top |
| 1822 | * group instead of the bottom group. |
| 1823 | */ |
| 1824 | if (collinear) { |
| 1825 | for (k = 0; noteptrs[k].top_p != 0; k++) { |
| 1826 | if (noteptrs[k].top_p->note_has_paren == YES) |
| 1827 | noteptrs[k].top_p->wlparen += 2 * halfwide |
| 1828 | - W_NORMAL * POINT; |
| 1829 | if (noteptrs[k].top_p->accidental != '\0') |
| 1830 | noteptrs[k].top_p->waccr += 2 * halfwide |
| 1831 | - W_NORMAL * POINT; |
| 1832 | } |
| 1833 | } |
| 1834 | } |
| 1835 | \f |
| 1836 | /* |
| 1837 | * Name: nextacc() |
| 1838 | * |
| 1839 | * Abstract: Find the next note that has an accidental to be processed. |
| 1840 | * |
| 1841 | * Returns: Index to the NOTE, or -1 if no more. |
| 1842 | * |
| 1843 | * Description: This function is called by doacc(), to return in the correct |
| 1844 | * order the notes that have accidentals to be processed. |
| 1845 | * (Actually, a note is to be processed not only if it has an |
| 1846 | * accidental, but also if it has parentheses.) The first time in |
| 1847 | * here, count is 0, and it looks for the first eligible note (top |
| 1848 | * down). The next time, count is 1, and it looks for the bottom- |
| 1849 | * most eligible note. After that, it goes through the inner |
| 1850 | * notes, top down. In the great majority of cases, this will |
| 1851 | * result in the most desirable packing of accidentals. |
| 1852 | */ |
| 1853 | |
| 1854 | static int |
| 1855 | nextacc(noteptrs, found) |
| 1856 | |
| 1857 | struct NOTEPTRS noteptrs[]; /* array of ptrs to notes to process */ |
| 1858 | int found; /* no. of accidentals found already */ |
| 1859 | |
| 1860 | { |
| 1861 | struct NOTE *note_p; /* point at a note */ |
| 1862 | static int previdx; /* idx to note chosen the last time in here */ |
| 1863 | static int lastidx; /* idx to the bottommost note chosen */ |
| 1864 | int n; /* loop counter */ |
| 1865 | |
| 1866 | |
| 1867 | /* |
| 1868 | * If this is the first call for this group(s), find the topmost |
| 1869 | * eligible note. |
| 1870 | */ |
| 1871 | if (found == 0) { |
| 1872 | for (n = 0; (note_p = GETPTR(n)) != 0; n++) { |
| 1873 | if (note_p->accidental != '\0' || |
| 1874 | note_p->note_has_paren == YES) { |
| 1875 | previdx = n; /* remember it for next time */ |
| 1876 | return (n); |
| 1877 | } |
| 1878 | } |
| 1879 | return (-1); /* no notes have acc or parens */ |
| 1880 | } |
| 1881 | |
| 1882 | /* |
| 1883 | * If this is the second call, find the bottom of the list, then look |
| 1884 | * backwards for the last eligible note. Stop before finding the first |
| 1885 | * note again. |
| 1886 | */ |
| 1887 | if (found == 1) { |
| 1888 | /* find the slot beyond the last note */ |
| 1889 | for (n = 0; (note_p = GETPTR(n)) != 0; n++) { |
| 1890 | ; |
| 1891 | } |
| 1892 | /* search from last note going backwards */ |
| 1893 | for (n-- ; n > previdx; n--) { |
| 1894 | note_p = GETPTR(n); |
| 1895 | if (note_p->accidental != '\0' || |
| 1896 | note_p->note_has_paren == YES) { |
| 1897 | lastidx = n; /* remember it for next time */ |
| 1898 | return (n); |
| 1899 | } |
| 1900 | } |
| 1901 | return (-1); /* only 1 note has acc or parens */ |
| 1902 | } |
| 1903 | |
| 1904 | /* |
| 1905 | * Third or later call: Scan inner notes top to bottom. |
| 1906 | */ |
| 1907 | for (n = previdx + 1; n < lastidx; n++) { |
| 1908 | note_p = GETPTR(n); |
| 1909 | if (note_p->accidental != '\0' || |
| 1910 | note_p->note_has_paren == YES) { |
| 1911 | previdx = n; |
| 1912 | return (n); |
| 1913 | } |
| 1914 | } |
| 1915 | return (-1); /* all eligible notes were already found */ |
| 1916 | } |
| 1917 | \f |
| 1918 | /* |
| 1919 | * Name: dodot() |
| 1920 | * |
| 1921 | * Abstract: Finds horizontal and vertical positions of dots. |
| 1922 | * |
| 1923 | * Returns: void |
| 1924 | * |
| 1925 | * Description: This function figures out the limitations on where dots |
| 1926 | * can be put, for each group, and calls dogrpdot() for each |
| 1927 | * group that has dots, to figure their positions. |
| 1928 | */ |
| 1929 | |
| 1930 | static void |
| 1931 | dodot(staff_p, gs1_p, gs2_p, halfwide, collinear) |
| 1932 | |
| 1933 | struct STAFF *staff_p; /* the staff the groups are connected to */ |
| 1934 | register struct GRPSYL *gs1_p, *gs2_p; /* point at group(s) in this hand */ |
| 1935 | double halfwide; /* half of max of width of notes */ |
| 1936 | int collinear; /* are stems collinear? */ |
| 1937 | |
| 1938 | { |
| 1939 | /* the highest and lowest values of steps above the middle staff */ |
| 1940 | /* line that a dot is allowed to be for the given group */ |
| 1941 | int uppermost, lowermost; |
| 1942 | |
| 1943 | int lowtopidx; /* index to lowest note of top group */ |
| 1944 | int push; /* steps to protruding note */ |
| 1945 | register int k; /* loop variable */ |
| 1946 | |
| 1947 | |
| 1948 | lowtopidx = gs1_p->nnotes - 1; /* for convenience */ |
| 1949 | |
| 1950 | /* |
| 1951 | * For each group that needs dots, set the outer limits of where |
| 1952 | * they are allowed. If the other group doesn't need dots, we |
| 1953 | * have to be careful to keep them out of its way. Otherwise, |
| 1954 | * don't worry about that; let them fall on top of each other if |
| 1955 | * that would happen. |
| 1956 | */ |
| 1957 | |
| 1958 | /* |
| 1959 | * If the first group needs dots, find out how high and low they are |
| 1960 | * allowed to be. Also find out if nearby notes in the other group |
| 1961 | * could be in the way of dots. Call dogrpdot() with this info to |
| 1962 | * find their positions. |
| 1963 | */ |
| 1964 | if (gs1_p->dots > 0) { |
| 1965 | /* upper limit is always as described above */ |
| 1966 | uppermost = gs1_p->notelist[0].stepsup; |
| 1967 | if (uppermost % 2 == 0) /* line note */ |
| 1968 | uppermost++; |
| 1969 | |
| 1970 | /* set lower limit as if no other group */ |
| 1971 | lowermost = gs1_p->notelist[lowtopidx].stepsup; |
| 1972 | if (lowermost % 2 == 0) /* line note */ |
| 1973 | lowermost--; |
| 1974 | |
| 1975 | /* but adjust if the other group exists & would interfere */ |
| 1976 | if (gs2_p != 0 && gs2_p->dots == 0 || collinear) { |
| 1977 | if (lowermost <= gs2_p->notelist[0].stepsup) |
| 1978 | lowermost += 2; |
| 1979 | } |
| 1980 | |
| 1981 | /* |
| 1982 | * If the stems are collinear, bottom group notes that are |
| 1983 | * in normal position for that group protrude to the right |
| 1984 | * relative to the top group. From top down, search for notes |
| 1985 | * in the bottom group that are like this. Set push to the |
| 1986 | * first one. If none are found, let push be 1000 to be out of |
| 1987 | * the way. In setting horizontal dot positions, dogrpdot() |
| 1988 | * needs to know this. |
| 1989 | */ |
| 1990 | push = 1000; |
| 1991 | if ( gs2_p != 0 && collinear ) { |
| 1992 | for (k = 0; k < gs2_p->nnotes; k++) { |
| 1993 | if (gs2_p->notelist[k].c[RX] == 0) { |
| 1994 | push = gs2_p->notelist[k].stepsup; |
| 1995 | break; |
| 1996 | } |
| 1997 | } |
| 1998 | } |
| 1999 | |
| 2000 | /* do top group's dots */ |
| 2001 | dogrpdot(staff_p, gs1_p, (struct GRPSYL *)0, halfwide, |
| 2002 | uppermost, lowermost, push); |
| 2003 | } |
| 2004 | |
| 2005 | /* |
| 2006 | * If the second group exists and needs dots, find out how high and |
| 2007 | * low they are allowed to be, and find their positions. |
| 2008 | */ |
| 2009 | if (gs2_p != 0 && gs2_p->dots > 0) { |
| 2010 | /* set upper limit as if no other group */ |
| 2011 | uppermost = gs2_p->notelist[0].stepsup; |
| 2012 | if (uppermost % 2 == 0) /* line note */ |
| 2013 | uppermost++; |
| 2014 | |
| 2015 | /* but adjust if the other group would interfere */ |
| 2016 | if (gs1_p->dots == 0 || collinear) { |
| 2017 | if (uppermost >= gs1_p->notelist[lowtopidx].stepsup) |
| 2018 | uppermost -= 2; |
| 2019 | } |
| 2020 | |
| 2021 | /* lower limit is always as described above */ |
| 2022 | lowermost = gs2_p->notelist[ gs2_p->nnotes - 1 ].stepsup; |
| 2023 | if (lowermost % 2 == 0) /* line note */ |
| 2024 | lowermost--; |
| 2025 | |
| 2026 | /* |
| 2027 | * Unless the stems are collinear, in which case no problem, |
| 2028 | * from bottom up, search for notes in the top group that |
| 2029 | * protrude towards the right. Set push to the first one. |
| 2030 | * If none are found, let push be 1000 to be out of the way. |
| 2031 | * In setting horizontal dot positions, dogrpdot() needs to |
| 2032 | * know this. |
| 2033 | */ |
| 2034 | push = 1000; |
| 2035 | if ( ! collinear ) { |
| 2036 | for (k = lowtopidx; k >= 0; k--) { |
| 2037 | if (gs1_p->notelist[k].c[RX] > 0) { |
| 2038 | push = gs1_p->notelist[k].stepsup; |
| 2039 | break; |
| 2040 | } |
| 2041 | } |
| 2042 | } |
| 2043 | |
| 2044 | /* do bottom group's dots */ |
| 2045 | dogrpdot(staff_p, gs2_p, gs1_p, halfwide, uppermost, lowermost, |
| 2046 | push); |
| 2047 | } |
| 2048 | } |
| 2049 | \f |
| 2050 | /* |
| 2051 | * Name: dogrpdot() |
| 2052 | * |
| 2053 | * Abstract: Finds horizontal and vertical positions of dots for one group. |
| 2054 | * |
| 2055 | * Returns: void |
| 2056 | * |
| 2057 | * Description: This function loops through all the notes belonging to the |
| 2058 | * given group, setting the coords of the dots relative to it. |
| 2059 | */ |
| 2060 | |
| 2061 | /* recover dotsteps from ydotr, avoiding roundoff error */ |
| 2062 | #define DOTSTEPS(ydotr) ( \ |
| 2063 | ydotr > 0.0 ? \ |
| 2064 | (int)((ydotr + 0.001) / STEPSIZE) \ |
| 2065 | : \ |
| 2066 | -(int)((-ydotr + 0.001) / STEPSIZE) \ |
| 2067 | ) |
| 2068 | |
| 2069 | static void |
| 2070 | dogrpdot(staff_p, gs_p, ogs_p, halfwide, uppermost, lowermost, push) |
| 2071 | |
| 2072 | struct STAFF *staff_p; /* the staff the groups are connected to */ |
| 2073 | register struct GRPSYL *gs_p; /* point at group */ |
| 2074 | struct GRPSYL *ogs_p; /* if we're doing group 1 and 2 together, and |
| 2075 | * gs_p is group 2, ogs_p is group 1, else 0 */ |
| 2076 | double halfwide; /* half of max of width of notes */ |
| 2077 | int uppermost; /* highest step where a dot is permitted */ |
| 2078 | int lowermost; /* lowest step where a dot is permitted */ |
| 2079 | int push; /* avoid protruding note at this position */ |
| 2080 | |
| 2081 | { |
| 2082 | float dotwidth; /* width of a dot (includes padding) */ |
| 2083 | int normhorz; /* use normal horizontal dot position? */ |
| 2084 | int notesteps; /* steps note is above center line of staff */ |
| 2085 | int dotsteps; /* steps dot is above center line of staff */ |
| 2086 | register int n, k; /* loop variables */ |
| 2087 | |
| 2088 | |
| 2089 | /* until proven otherwise, assume normal horizontal dot position */ |
| 2090 | normhorz = YES; |
| 2091 | |
| 2092 | /* |
| 2093 | * The rules for vertical positioning of dots are as follows. |
| 2094 | * For space notes, dots will be put in the same space. For line |
| 2095 | * notes we'd like them to be in the space directly above, except for |
| 2096 | * voice 2 in vscheme=2o,3o or 2f,3f when voice 1 is not space, in |
| 2097 | * which case we'd like them to be in the space below. But if notes in |
| 2098 | * a group are jammed onto neighboring steps, we may need to put some |
| 2099 | * line note dots on the space below regardless; and we may |
| 2100 | * even have to let some dots land on top of each other. But in |
| 2101 | * any case, never exceed the uppermost/lowermost bounds, which |
| 2102 | * would interfere with the other group. |
| 2103 | * |
| 2104 | * The rules for horizontal positioning of dots are as follows. |
| 2105 | * If the note on the dot's space, or either neighboring line, |
| 2106 | * is in abnormal position to the right, the dot must be put |
| 2107 | * farther right than normal. The parameter "push" is the nearest |
| 2108 | * note from the other group that protrudes this way. And the dots |
| 2109 | * of all the notes have to line up, so if any one has this problem, |
| 2110 | * they must all be moved. |
| 2111 | */ |
| 2112 | |
| 2113 | /* |
| 2114 | * Loop through all notes in the group, setting dot positions. At |
| 2115 | * the top of the loop, "dotsteps" is the previous dot, but by the |
| 2116 | * end it gets set to the current dot. |
| 2117 | */ |
| 2118 | dotsteps = uppermost + 2; /* pretend previous dot was here */ |
| 2119 | |
| 2120 | for (n = 0; n < gs_p->nnotes; n++) { |
| 2121 | |
| 2122 | notesteps = gs_p->notelist[n].stepsup; |
| 2123 | |
| 2124 | if (notesteps % 2 == 0) { |
| 2125 | /* |
| 2126 | * This note is on a line. If the dot cannot be put |
| 2127 | * above the line, or if doing that would overlay the |
| 2128 | * previous dot and we are allowed to put it below |
| 2129 | * the line, then put it below the line. Else, put |
| 2130 | * it above the line. Notice that we're putting the |
| 2131 | * dot in the space above if at all possible; later on, |
| 2132 | * we'll make adjustments for voice 2 if appropriate. |
| 2133 | */ |
| 2134 | if (notesteps + 1 > uppermost || |
| 2135 | (notesteps + 1 == dotsteps && |
| 2136 | notesteps - 1 >= lowermost)) { |
| 2137 | dotsteps = notesteps - 1; |
| 2138 | } else { |
| 2139 | dotsteps = notesteps + 1; |
| 2140 | } |
| 2141 | } else { |
| 2142 | /* |
| 2143 | * This note is on a space. The dot must be put in |
| 2144 | * this same space, regardless of anything else. |
| 2145 | */ |
| 2146 | dotsteps = notesteps; |
| 2147 | } |
| 2148 | |
| 2149 | /* set relative y coord based on step position */ |
| 2150 | gs_p->notelist[n].ydotr = dotsteps * STEPSIZE; |
| 2151 | |
| 2152 | /* |
| 2153 | * Now see if this dot forces abnormal positioning. "Push" may |
| 2154 | * indicate a protruding note in the other group. If this |
| 2155 | * note is within 1 step of our dot, use abnormal positioning |
| 2156 | * for the dot. Else if the stem is down, all dots can be |
| 2157 | * normal. Else, we have to search for protruding notes to |
| 2158 | * see where the dot can be. |
| 2159 | */ |
| 2160 | if (normhorz == YES) { |
| 2161 | if (abs(dotsteps - push) <= 1) { |
| 2162 | normhorz = NO; |
| 2163 | } else if (gs_p->stemdir == UP) { |
| 2164 | for (k = 0; k < gs_p->nnotes; k++) { |
| 2165 | notesteps = gs_p->notelist[k].stepsup; |
| 2166 | |
| 2167 | if (gs_p->notelist[k].c[RE] >halfwide && |
| 2168 | notesteps <= dotsteps + 1 && |
| 2169 | notesteps >= dotsteps - 1) { |
| 2170 | |
| 2171 | normhorz = NO; |
| 2172 | break; |
| 2173 | } |
| 2174 | } |
| 2175 | } |
| 2176 | } |
| 2177 | } |
| 2178 | |
| 2179 | /* |
| 2180 | * Set horizontal dot positions, relative to the group. STDPAD is |
| 2181 | * needed because notehead characters don't include padding. The |
| 2182 | * abnormal case adds in one more notehead width, minus the width |
| 2183 | * of the stem. Since the dots for all notes line up vertically, |
| 2184 | * xdotr is in GRPSYL instead of in each NOTE. |
| 2185 | */ |
| 2186 | dotwidth = width(FONT_MUSIC, DFLT_SIZE, C_DOT); |
| 2187 | gs_p->xdotr = halfwide + STDPAD + dotwidth / 2; |
| 2188 | if (normhorz == NO) { |
| 2189 | gs_p->xdotr += 2 * halfwide - W_NORMAL * POINT; |
| 2190 | } |
| 2191 | |
| 2192 | /* |
| 2193 | * If this is voice 2, we may need to adjust the vertical position of |
| 2194 | * nonshared line notes. The same should happen if this is voice 3 |
| 2195 | * "standing in" for voice 2. |
| 2196 | */ |
| 2197 | if (gs_p->pvno == 2) { |
| 2198 | int trymove; /* try to move dots? */ |
| 2199 | int vscheme; /* voice scheme */ |
| 2200 | RATIONAL vtime; /* time so far in this measure */ |
| 2201 | int prevdotsteps; /* Y distance of prev note's dot */ |
| 2202 | struct GRPSYL *pgs_p; /* point along GRPSYL list */ |
| 2203 | int onotesteps; /* lowest note of voice 1 */ |
| 2204 | |
| 2205 | trymove = NO; /* first assume leave them alone */ |
| 2206 | vscheme = svpath(gs_p->staffno, VSCHEME)->vscheme; |
| 2207 | if (vscheme == V_2OPSTEM || vscheme == V_3OPSTEM) { |
| 2208 | /* always try to move if 2o or 3o */ |
| 2209 | trymove = YES; |
| 2210 | } else { |
| 2211 | /* 2f or 3f; move iff voice 1 is not all spaces here */ |
| 2212 | vtime = Zero; /* add up time of preceding groups */ |
| 2213 | for (pgs_p = gs_p->prev; pgs_p != 0; |
| 2214 | pgs_p = pgs_p->prev) { |
| 2215 | vtime = radd(vtime, pgs_p->fulltime); |
| 2216 | } |
| 2217 | if ( ! hasspace(staff_p->groups_p[0], vtime, |
| 2218 | radd(vtime, gs_p->fulltime))) { |
| 2219 | /* not all space during duration of our group*/ |
| 2220 | trymove = YES; |
| 2221 | } |
| 2222 | } |
| 2223 | |
| 2224 | if (trymove == YES) { |
| 2225 | /* |
| 2226 | * We need to try to move the dots of line notes from |
| 2227 | * the space above them to the space below them. We |
| 2228 | * will work from bottom to top. Initially, pretend |
| 2229 | * that the previous note is way low out of the way. |
| 2230 | * If a voice 1 group was being handled along with our |
| 2231 | * group, find the stepsup of its lowest note. |
| 2232 | */ |
| 2233 | prevdotsteps = -1000; |
| 2234 | if (ogs_p != 0) { |
| 2235 | onotesteps = ogs_p->notelist[ |
| 2236 | ogs_p->nnotes - 1].stepsup; |
| 2237 | } else { |
| 2238 | onotesteps = 0; /* for lint; set before used */ |
| 2239 | } |
| 2240 | for (n = gs_p->nnotes - 1; n >= 0; n--) { |
| 2241 | notesteps = gs_p->notelist[n].stepsup; |
| 2242 | /* |
| 2243 | * We want to stop if we run into notes shared |
| 2244 | * by group 1 if it exists. ( > is defensive). |
| 2245 | */ |
| 2246 | if (ogs_p != 0 && notesteps >= onotesteps) |
| 2247 | break; |
| 2248 | /* |
| 2249 | * Recover our dotsteps from our dots coord |
| 2250 | * calculated earlier in this function. Then, |
| 2251 | * consider moving our dot only if we are a |
| 2252 | * line note and our dot is currently in the |
| 2253 | * space above. (It could already be below, |
| 2254 | * do to tightly packed notes.) |
| 2255 | */ |
| 2256 | dotsteps = DOTSTEPS(gs_p->notelist[n].ydotr); |
| 2257 | if (notesteps % 2 == 0 && |
| 2258 | dotsteps == notesteps + 1) { |
| 2259 | /* |
| 2260 | * If the previous (lower) note is at |
| 2261 | * least 2 steps away, we can certainly |
| 2262 | * move our dot. But also move it if |
| 2263 | * we are the top note of group 2, and |
| 2264 | * group 1 exists and has a note 2 steps |
| 2265 | * away, and they don't have a dot at |
| 2266 | * the same horz position; because our |
| 2267 | * dot would be confusing if above. If |
| 2268 | * it make our dot land on top of the |
| 2269 | * previous note's dot, tough. |
| 2270 | */ |
| 2271 | if (prevdotsteps < notesteps - 1 || |
| 2272 | n == 0 && ogs_p != 0 && |
| 2273 | notesteps + 2 == onotesteps && |
| 2274 | ogs_p->xdotr != gs_p->xdotr) { |
| 2275 | |
| 2276 | dotsteps -= 2; |
| 2277 | gs_p->notelist[n].ydotr -= |
| 2278 | 2.0 * STEPSIZE; |
| 2279 | } |
| 2280 | } |
| 2281 | prevdotsteps = dotsteps; |
| 2282 | } |
| 2283 | } |
| 2284 | } |
| 2285 | } |
| 2286 | \f |
| 2287 | /* |
| 2288 | * Name: westwith() |
| 2289 | * |
| 2290 | * Abstract: Adjust west coord of a group to allow for its "with" lists. |
| 2291 | * |
| 2292 | * Returns: void |
| 2293 | * |
| 2294 | * Description: This function is given a GRPSYL whose relative horizontal |
| 2295 | * coords are set, relative to the center of the group, except |
| 2296 | * that "with" lists have not yet been considered. It alters |
| 2297 | * gs_p->c[RW] if need be so that the group's rectangle includes |
| 2298 | * all "with" lists. |
| 2299 | */ |
| 2300 | |
| 2301 | static void |
| 2302 | westwith(gs_p) |
| 2303 | |
| 2304 | struct GRPSYL *gs_p; /* point at this group */ |
| 2305 | |
| 2306 | { |
| 2307 | int n; /* loop through the "with" list items */ |
| 2308 | int font, size; /* of the chars in the "with" list item */ |
| 2309 | int first_char; /* first char of string to print */ |
| 2310 | char *str_p; /* point into the item */ |
| 2311 | float x_offset; /* half the width of the first char in item */ |
| 2312 | |
| 2313 | |
| 2314 | for (n = 0; n < gs_p->nwith; n++) { |
| 2315 | /* should center first character on x */ |
| 2316 | font = gs_p->withlist[n][0]; |
| 2317 | size = gs_p->withlist[n][1]; |
| 2318 | str_p = gs_p->withlist[n] + 2; |
| 2319 | first_char = next_str_char(&str_p, &font, &size); |
| 2320 | x_offset = width(font, size, first_char) / 2.0; |
| 2321 | if (-x_offset < gs_p->c[RW]) |
| 2322 | gs_p->c[RW] = -x_offset; |
| 2323 | } |
| 2324 | } |
| 2325 | \f |
| 2326 | /* |
| 2327 | * Name: eastwith() |
| 2328 | * |
| 2329 | * Abstract: Adjust east coord of a group to allow for its "with" lists. |
| 2330 | * |
| 2331 | * Returns: void |
| 2332 | * |
| 2333 | * Description: This function is given a GRPSYL whose relative horizontal |
| 2334 | * coords are set, relative to the center of the group, except |
| 2335 | * that "with" lists have not yet been considered. It alters |
| 2336 | * gs_p->c[RE] if need be so that the group's rectangle includes |
| 2337 | * all "with" lists. |
| 2338 | */ |
| 2339 | |
| 2340 | static void |
| 2341 | eastwith(gs_p) |
| 2342 | |
| 2343 | struct GRPSYL *gs_p; /* point at this group */ |
| 2344 | |
| 2345 | { |
| 2346 | int n; /* loop through the "with" list items */ |
| 2347 | int font, size; /* of the chars in the "with" list item */ |
| 2348 | int first_char; /* first char of string to print */ |
| 2349 | char *str_p; /* point into the item */ |
| 2350 | float x_offset; /* half the width of the first char in item */ |
| 2351 | |
| 2352 | |
| 2353 | for (n = 0; n < gs_p->nwith; n++) { |
| 2354 | /* should center first character on x */ |
| 2355 | font = gs_p->withlist[n][0]; |
| 2356 | size = gs_p->withlist[n][1]; |
| 2357 | str_p = gs_p->withlist[n] + 2; |
| 2358 | first_char = next_str_char(&str_p, &font, &size); |
| 2359 | x_offset = strwidth(gs_p->withlist[n]) - |
| 2360 | width(font, size, first_char) / 2.0; |
| 2361 | if (x_offset > gs_p->c[RE]) |
| 2362 | gs_p->c[RE] = x_offset; |
| 2363 | } |
| 2364 | } |
| 2365 | \f |
| 2366 | /* |
| 2367 | * Name: csbstempad() |
| 2368 | * |
| 2369 | * Abstract: Pad a group's RW for cross staff beaming if need be. |
| 2370 | * |
| 2371 | * Returns: void |
| 2372 | * |
| 2373 | * Description: In cross staff beamed groups, where the beams are between the |
| 2374 | * staffs, and a note on the bottom staff is followed by a note on |
| 2375 | * the top staff, and the first note has no dots or anything else |
| 2376 | * that would force more space after it, and the top note has no |
| 2377 | * accidentals, graces, or anything that would force more space |
| 2378 | * before it, the stems of the two groups can be very close |
| 2379 | * together, too close. This function checks for that case, and |
| 2380 | * when found, adds padding to the left of the top group. |
| 2381 | */ |
| 2382 | |
| 2383 | static void |
| 2384 | csbstempad(mll_p, gs_p) |
| 2385 | |
| 2386 | struct MAINLL *mll_p; /* the MLL item the group is connected to */ |
| 2387 | struct GRPSYL *gs_p; /* point at the top staff's group */ |
| 2388 | |
| 2389 | { |
| 2390 | struct GRPSYL *gs2_p; /* point at various GRPSYLs */ |
| 2391 | struct CHORD *ch_p, *pch_p; /* our chord and preceding chord */ |
| 2392 | struct MAINLL *m2_p; /* loop through MLL */ |
| 2393 | int k; /* loop through notelist */ |
| 2394 | int found; /* have we found our group? */ |
| 2395 | |
| 2396 | |
| 2397 | /* if this group is not a candidate for this, return */ |
| 2398 | if (gs_p->beamto != CS_BELOW) /* must be CSB beamed with below */ |
| 2399 | return; |
| 2400 | if (gs_p->stemdir == UP) /* stem must be down */ |
| 2401 | return; |
| 2402 | if (gs_p->beamloc == STARTITEM) /* must not be first item in CSB */ |
| 2403 | return; |
| 2404 | if (gs_p->prev == 0) /* (defensive) */ |
| 2405 | return; |
| 2406 | if (gs_p->prev->grpcont != GC_SPACE) /* prev must be a space */ |
| 2407 | return; |
| 2408 | |
| 2409 | /* |
| 2410 | * The notes should all have the same RW (even cues) unless a note is |
| 2411 | * on the "wrong" side of the stem, because they are all supposed to |
| 2412 | * touch the stem. In the latter case, there's already enough space in |
| 2413 | * the group to the left of the stem, so return. |
| 2414 | */ |
| 2415 | for (k = 1; k < gs_p->nnotes; k++) { |
| 2416 | if (ABSDIFF(gs_p->notelist[k].c[RW], gs_p->notelist[0].c[RW]) |
| 2417 | > FUDGE) |
| 2418 | return; |
| 2419 | } |
| 2420 | |
| 2421 | /* |
| 2422 | * If there's anything to the left of the notes' RWs (the stem |
| 2423 | * position), it should be enough space, so return. |
| 2424 | */ |
| 2425 | if (gs_p->c[RW] < gs_p->notelist[0].c[RW] - STDPAD - FUDGE) |
| 2426 | return; |
| 2427 | |
| 2428 | /* find the chord headcell for this measure */ |
| 2429 | for (m2_p = mll_p->prev; m2_p->str != S_CHHEAD; m2_p = m2_p->prev) |
| 2430 | ; |
| 2431 | /* |
| 2432 | * Loop through the chords. For each chord, loop through all its |
| 2433 | * groups, trying to find our group. It should be found. At the point |
| 2434 | * it is found, pch_p will point to the chord preceding the one that |
| 2435 | * contains our group. |
| 2436 | */ |
| 2437 | found = NO; |
| 2438 | pch_p = 0; /* to avoid useless 'used before set' warning */ |
| 2439 | for (ch_p = m2_p->u.chhead_p->ch_p; ch_p != 0; |
| 2440 | pch_p = ch_p, ch_p = ch_p->ch_p) { |
| 2441 | for (gs2_p = ch_p->gs_p; gs2_p != 0; gs2_p = gs2_p->gs_p) { |
| 2442 | if (gs2_p == gs_p) { |
| 2443 | found = YES; |
| 2444 | break; |
| 2445 | } |
| 2446 | } |
| 2447 | if (found == YES) |
| 2448 | break; |
| 2449 | } |
| 2450 | if (found == NO) /* defensive; this should never happen */ |
| 2451 | return; |
| 2452 | |
| 2453 | /* find next visible staff after our staff */ |
| 2454 | for (m2_p = mll_p->next; m2_p->str == S_STAFF && |
| 2455 | m2_p->u.staff_p->visible == NO; m2_p = m2_p->next) |
| 2456 | ; |
| 2457 | if (m2_p->str != S_STAFF) /* defensive; should not happen */ |
| 2458 | return; |
| 2459 | |
| 2460 | /* |
| 2461 | * Loop down the preceding chord, looking for a group that is on the |
| 2462 | * next visible staff after our staff and is CSB'ed to the staff above. |
| 2463 | */ |
| 2464 | for (gs2_p = pch_p->gs_p; gs2_p != 0; gs2_p = gs2_p->gs_p) { |
| 2465 | |
| 2466 | if (gs2_p->staffno == m2_p->u.staff_p->staffno && |
| 2467 | gs2_p->beamto == CS_ABOVE) { |
| 2468 | /* |
| 2469 | * We found such a group; it must be the only one. |
| 2470 | * Check that it meets the conditions. |
| 2471 | */ |
| 2472 | if (gs2_p->stemdir == DOWN) |
| 2473 | return; |
| 2474 | /* |
| 2475 | * The notes need to all have the same RE, analogous to |
| 2476 | * the earlier check on gs_p's RW. |
| 2477 | */ |
| 2478 | for (k = 1; k < gs2_p->nnotes; k++) { |
| 2479 | if (ABSDIFF(gs2_p->notelist[k].c[RE], gs2_p-> |
| 2480 | notelist[0].c[RE]) > FUDGE) |
| 2481 | return; |
| 2482 | } |
| 2483 | /* |
| 2484 | * If there's anything to the right of the notes' REs, |
| 2485 | * there's already enough space. |
| 2486 | */ |
| 2487 | if (gs2_p->c[RE] > gs2_p->notelist[0].c[RE] + |
| 2488 | STDPAD + FUDGE) |
| 2489 | return; |
| 2490 | |
| 2491 | /* |
| 2492 | * FINALLY! We have established the need for more |
| 2493 | * space. Append it to our group's RW. |
| 2494 | */ |
| 2495 | gs_p->c[RW] -= STEPSIZE; |
| 2496 | return; |
| 2497 | } |
| 2498 | } |
| 2499 | |
| 2500 | /* didn't find one; shouldn't happen, but just return */ |
| 2501 | } |
| 2502 | \f |
| 2503 | /* |
| 2504 | * Name: proctab() |
| 2505 | * |
| 2506 | * Abstract: Sets relative horizontal coords of fret numbers. |
| 2507 | * |
| 2508 | * Returns: void |
| 2509 | * |
| 2510 | * Description: This function sets all the horizontal coords of "notes" on a |
| 2511 | * tablature staff, which are actually fret numbers. It sets RW |
| 2512 | * and RE for the group, too. They also take bends into account. |
| 2513 | */ |
| 2514 | |
| 2515 | static void |
| 2516 | proctab(mll_p, staff_p, gs_p) |
| 2517 | |
| 2518 | struct MAINLL *mll_p; /* the MLL item the group is connected to */ |
| 2519 | struct STAFF *staff_p; /* the staff the group is connected to */ |
| 2520 | struct GRPSYL *gs_p; /* point at this group */ |
| 2521 | |
| 2522 | { |
| 2523 | int n; /* loop through the "notes" in the group */ |
| 2524 | float halfwide; /* half the width of a fret or bend number */ |
| 2525 | float maxhalffret; /* half the max width of a fret number */ |
| 2526 | float maxhalfbend; /* half the max width of a bend number */ |
| 2527 | float maxbend; /* width of a bend number that sticks right */ |
| 2528 | struct GRPSYL *prevgs_p;/* point at previous group */ |
| 2529 | int center; /* should bend string be centered? */ |
| 2530 | int k; /* loop variable */ |
| 2531 | |
| 2532 | |
| 2533 | maxhalffret = 0.0; |
| 2534 | maxhalfbend = 0.0; |
| 2535 | maxbend = 0.0; |
| 2536 | |
| 2537 | prevgs_p = prevgrpsyl(gs_p, &mll_p); /* in case we need it */ |
| 2538 | |
| 2539 | /* loop though all frets and bends in this group */ |
| 2540 | for (n = 0; n < gs_p->nnotes; n++) { |
| 2541 | /* |
| 2542 | * If there is a fret, find half the width of that number. It |
| 2543 | * should be centered on the center of the group. Keep track |
| 2544 | * of the maximum width so far. Allow 1.5*STDPAD on each side |
| 2545 | * of the fret number, since we don't ever want the numbers so |
| 2546 | * close that they look like one number. |
| 2547 | */ |
| 2548 | if (gs_p->notelist[n].FRETNO != NOFRET) { |
| 2549 | halfwide = strwidth(fret_string(&gs_p->notelist[n], |
| 2550 | gs_p)) / 2.0; |
| 2551 | gs_p->notelist[n].c[RX] = 0.0; |
| 2552 | gs_p->notelist[n].c[RE] = halfwide; |
| 2553 | gs_p->notelist[n].c[RW] = -halfwide; |
| 2554 | maxhalffret = MAX(halfwide + 1.5*STDPAD, maxhalffret); |
| 2555 | } |
| 2556 | |
| 2557 | /* |
| 2558 | * If there is a bend, figure out if it's the normal situation |
| 2559 | * (centered on the group's X) or the the case where its left |
| 2560 | * edge should be at the group's X (the case of a continuation |
| 2561 | * bend where the previous group's bend was higher). In the |
| 2562 | * latter case, the string had to be shifted to avoid colliding |
| 2563 | * with the arrow coming down from the previous group. |
| 2564 | */ |
| 2565 | if (HASREALBEND(gs_p->notelist[n])) { |
| 2566 | center = YES; /* first assume normal */ |
| 2567 | |
| 2568 | /* search previous group, if any, for a bend */ |
| 2569 | if (prevgs_p != 0) { |
| 2570 | for (k = 0; k < prevgs_p->nnotes; k++) { |
| 2571 | if (HASREALBEND(prevgs_p->notelist[k])) |
| 2572 | break; |
| 2573 | } |
| 2574 | /* |
| 2575 | * If previous group had a bend and its |
| 2576 | * distance was higher than the current group, |
| 2577 | * we have the special case. |
| 2578 | */ |
| 2579 | if (k < prevgs_p->nnotes && |
| 2580 | GT( ratbend(&prevgs_p->notelist[k]), |
| 2581 | ratbend(&gs_p->notelist[n]) ) ) { |
| 2582 | center = NO; |
| 2583 | } |
| 2584 | } |
| 2585 | if (center == YES) { |
| 2586 | /* |
| 2587 | * Normal case of a bend string: centered at |
| 2588 | * group's X. Maintain maxhalfbend as the |
| 2589 | * the widest so far. |
| 2590 | */ |
| 2591 | halfwide = strwidth(bend_string( |
| 2592 | &gs_p->notelist[n])) / 2.0; |
| 2593 | maxhalfbend = MAX(halfwide, maxhalfbend); |
| 2594 | } else { |
| 2595 | /* |
| 2596 | * A bend string that has its left edge at the |
| 2597 | * group's X. There can only be one such, |
| 2598 | * since multiple continuation bends are not |
| 2599 | * allowed (other than releases). |
| 2600 | */ |
| 2601 | maxbend = strwidth(bend_string( |
| 2602 | &gs_p->notelist[n])); |
| 2603 | } |
| 2604 | } |
| 2605 | } |
| 2606 | |
| 2607 | /* |
| 2608 | * Set the group's relative horizontal coordinates. On the east, add |
| 2609 | * extra room if there are ties or slurs. On the west, add any user |
| 2610 | * requested padding. Also adjust for "with" lists. They can extend |
| 2611 | * into tie/slur padding, but not into user requested padding. |
| 2612 | */ |
| 2613 | gs_p->c[RX] = 0.0; |
| 2614 | |
| 2615 | gs_p->c[RW] = -MAX(maxhalffret, maxhalfbend); |
| 2616 | westwith(gs_p); |
| 2617 | gs_p->c[RW] -= gs_p->padding; |
| 2618 | gs_p->c[RW] -= vvpath(gs_p->staffno, gs_p->vno, PAD)->pad; |
| 2619 | |
| 2620 | maxhalffret += tieslurpad(staff_p, gs_p); |
| 2621 | gs_p->c[RE] = MAX(MAX(maxhalffret, maxhalfbend), maxbend); |
| 2622 | eastwith(gs_p); |
| 2623 | } |
| 2624 | \f |
| 2625 | /* |
| 2626 | * Name: noterparen() |
| 2627 | * |
| 2628 | * Abstract: Finds horizontal position notes' right parentheses. |
| 2629 | * |
| 2630 | * Returns: void |
| 2631 | * |
| 2632 | * Description: If any of the notes in the given group(s) are to have |
| 2633 | * parentheses around them, this function finds the horizontal |
| 2634 | * positions of the right parentheses. The left ones were done |
| 2635 | * in doacc() along with accidentals. For each group, it uses |
| 2636 | * the appropriate size of parentheses (based on normal versus |
| 2637 | * cue/grace), and places them appropriately, considering also |
| 2638 | * the size of the notes. However, if there are two groups, |
| 2639 | * the note head sizes could be different. The halfwide and |
| 2640 | * halfhigh passed in are supposed to be the right size for the |
| 2641 | * bigger of the two sizes, and accidentals will not be packed |
| 2642 | * as tightly against the other notes. This doesn't hurt, and |
| 2643 | * isn't worth the trouble to do it "right". |
| 2644 | */ |
| 2645 | |
| 2646 | static void |
| 2647 | noterparen(noteptrs, gs1_p, gs2_p, halfwide, halfhigh, collinear) |
| 2648 | |
| 2649 | struct NOTEPTRS noteptrs[]; /* array of ptrs to notes to process */ |
| 2650 | struct GRPSYL *gs1_p, *gs2_p; /* point at group(s) in this hand */ |
| 2651 | double halfwide; /* half of max of width & height of (notes */ |
| 2652 | double halfhigh; /* in group 1, notes in group 2) */ |
| 2653 | int collinear; /* are stems collinear? */ |
| 2654 | |
| 2655 | { |
| 2656 | /* |
| 2657 | * Each structure in this table represents either a note head that is |
| 2658 | * farther right than normal, note dot(s), or right paren. A note head |
| 2659 | * could be too far right for one of two reasons: either it was |
| 2660 | * forced to be on the right ("wrong") side of a stem that points |
| 2661 | * up, or it is a normal note in the bottom group when the stems are |
| 2662 | * collinear. In the collinear case, to make this function easier, |
| 2663 | * we start out regarding the top group as being normal, and |
| 2664 | * the bottom group as being shifted right one note head, and we figure |
| 2665 | * everything relative to the top group. But at the end we adjust |
| 2666 | * so that every parenthesis is relative to its own group, like |
| 2667 | * it's supposed to be. |
| 2668 | * |
| 2669 | * The coordinates define the rectangle that surrounds the note, dot(s), |
| 2670 | * or paren, including standard padding, even on note heads, which don't |
| 2671 | * normally have padding. First the notes and dots are put into this |
| 2672 | * table, just one rectangle for a sequence of dots; then the right |
| 2673 | * parens one at a time, making sure they don't overlap things already |
| 2674 | * in the table. |
| 2675 | * |
| 2676 | * To see if the parenthesis being added overlaps, first its north |
| 2677 | * and south are tested. All previous rectangles that are "out of |
| 2678 | * its way" vertically are marked not "relevant"; the others are |
| 2679 | * marked "relevant". As positions are tried, left to right, positions |
| 2680 | * that fail to avoid overlap are marked "tried". |
| 2681 | */ |
| 2682 | struct { |
| 2683 | float n, s, e, w; /* boundaries of a rectangle */ |
| 2684 | short relevant; /* is rectangle relevant? */ |
| 2685 | short tried; /* have we tried this one yet? */ |
| 2686 | } rectab[2 * MAXHAND + 1]; /* enough for all notes & accidentals*/ |
| 2687 | |
| 2688 | struct NOTE *note_p; /* point at a note */ |
| 2689 | int reclim; /* index after last rectangle in tab */ |
| 2690 | int parensexist; /* does any note have parens? */ |
| 2691 | float north, south, east, west; /* relative coords of new accidental */ |
| 2692 | float parenwidth; /* width of note's left parenthesis */ |
| 2693 | float parenv; /* half the vertical size of paren */ |
| 2694 | float dotoff; /* additional offset caused by dots */ |
| 2695 | float dotoff1, dotoff2; /* same, for groups 1 and 2 */ |
| 2696 | int overlap; /* does our acc overlap existing ones*/ |
| 2697 | int try; /* which element of rectab to try */ |
| 2698 | int k, j; /* loop variables */ |
| 2699 | int size; |
| 2700 | |
| 2701 | |
| 2702 | /* |
| 2703 | * If no notes have parentheses, we can get out because there is |
| 2704 | * nothing to do. |
| 2705 | */ |
| 2706 | parensexist = NO; /* init to no parens */ |
| 2707 | for (k = 0; (note_p = GETPTR(k)) != 0; k++) { |
| 2708 | if (note_p->note_has_paren == YES) |
| 2709 | parensexist = YES; |
| 2710 | } |
| 2711 | if (parensexist == NO) |
| 2712 | return; |
| 2713 | |
| 2714 | reclim = 0; /* table initially empty */ |
| 2715 | |
| 2716 | /* set up dot offsets for both groups, zero if no dots */ |
| 2717 | dotoff1 = gs1_p->dots * (width(FONT_MUSIC,DFLT_SIZE,C_DOT) + 2*STDPAD); |
| 2718 | dotoff2 = 0.0; /* prevent useless 'used before set' warning */ |
| 2719 | if (gs2_p != 0) { |
| 2720 | dotoff2 = gs2_p->dots * (width(FONT_MUSIC, DFLT_SIZE, C_DOT) + |
| 2721 | 2 * STDPAD); |
| 2722 | } |
| 2723 | |
| 2724 | /* |
| 2725 | * Loop through noteptrs, loading rectab with all the things that are |
| 2726 | * already present that are to the right of the baseline. |
| 2727 | */ |
| 2728 | for (k = 0; (note_p = GETPTR(k)) != 0; k++) { |
| 2729 | /* |
| 2730 | * If note exists in top group, use its dot offset, else use |
| 2731 | * bottom's. If it's in both, the results would be the same. |
| 2732 | */ |
| 2733 | if (noteptrs[k].top_p != 0) |
| 2734 | dotoff = dotoff1; |
| 2735 | else |
| 2736 | dotoff = dotoff2; |
| 2737 | |
| 2738 | /* if note is right of normal position, put it in the table */ |
| 2739 | if (note_p->c[RX] > 0) { |
| 2740 | rectab[reclim].n = note_p->c[RY] + halfhigh + STDPAD; |
| 2741 | rectab[reclim].s = note_p->c[RY] - halfhigh - STDPAD; |
| 2742 | rectab[reclim].e = note_p->c[RE] + STDPAD; |
| 2743 | rectab[reclim].w = note_p->c[RW] - STDPAD; |
| 2744 | reclim++; |
| 2745 | } |
| 2746 | |
| 2747 | /* if collinear, bottom group's notes go into table if normal */ |
| 2748 | if (collinear && noteptrs[k].bot_p != 0) { |
| 2749 | if (note_p->c[RX] == 0) { |
| 2750 | rectab[reclim].n = note_p->c[RY] + halfhigh |
| 2751 | + STDPAD; |
| 2752 | rectab[reclim].s = note_p->c[RY] - halfhigh |
| 2753 | - STDPAD; |
| 2754 | rectab[reclim].e = W_NORMAL * POINT |
| 2755 | + 3 * halfwide + STDPAD; |
| 2756 | rectab[reclim].w = W_NORMAL * POINT |
| 2757 | + halfwide - STDPAD; |
| 2758 | reclim++; |
| 2759 | } |
| 2760 | } |
| 2761 | |
| 2762 | /* if this group has dots, do rectangle for dots */ |
| 2763 | if (dotoff > 0) { |
| 2764 | rectab[reclim].n = note_p->ydotr + STDPAD; |
| 2765 | rectab[reclim].s = note_p->ydotr - STDPAD; |
| 2766 | if (noteptrs[k].top_p != 0) |
| 2767 | rectab[reclim].e = gs1_p->xdotr + dotoff; |
| 2768 | else |
| 2769 | rectab[reclim].e = gs2_p->xdotr + dotoff; |
| 2770 | rectab[reclim].w = 0; |
| 2771 | reclim++; |
| 2772 | } |
| 2773 | } |
| 2774 | |
| 2775 | /* |
| 2776 | * Loop through all parentheses, finding where they will fit, storing |
| 2777 | * that info in erparen, and adding them to rectab. |
| 2778 | */ |
| 2779 | for (k = 0; (note_p = GETPTR(k)) != 0; k++) { |
| 2780 | |
| 2781 | /* if no parens around the note, skip the note */ |
| 2782 | if (note_p->note_has_paren == NO) |
| 2783 | continue; |
| 2784 | |
| 2785 | /* get dimensions of note's right paren */ |
| 2786 | size = (note_p->notesize == GS_NORMAL ? DFLT_SIZE : SMALLSIZE); |
| 2787 | parenwidth = width(FONT_TR, size, ')'); |
| 2788 | parenv = height(FONT_TR, size, ')') / 2.0; |
| 2789 | |
| 2790 | /* set the north and south of the paren */ |
| 2791 | north = note_p->c[RY] + parenv; |
| 2792 | south = note_p->c[RY] - parenv; |
| 2793 | |
| 2794 | /* |
| 2795 | * For each rectangle in rectab, decide whether (based on |
| 2796 | * its vertical coords) it could possibly overlap with our |
| 2797 | * new paren. If it's totally above or below ours, it |
| 2798 | * can't. We allow a slight overlap (FUDGE) so that round |
| 2799 | * off errors don't stop us from packing things as tightly |
| 2800 | * as possible. |
| 2801 | */ |
| 2802 | for (j = 0; j < reclim; j++) { |
| 2803 | if (rectab[j].s + FUDGE > north || |
| 2804 | rectab[j].n < south + FUDGE) |
| 2805 | rectab[j].relevant = NO; |
| 2806 | else |
| 2807 | rectab[j].relevant = YES; |
| 2808 | } |
| 2809 | |
| 2810 | /* |
| 2811 | * Mark that none of the relevant rectangles' boundaries have |
| 2812 | * been tried yet for positioning our paren. |
| 2813 | */ |
| 2814 | for (j = 0; j < reclim; j++) { |
| 2815 | if (rectab[j].relevant == YES) |
| 2816 | rectab[j].tried = NO; |
| 2817 | } |
| 2818 | |
| 2819 | /* |
| 2820 | * Set up first trial position for this paren, just to the |
| 2821 | * right of normal notes, allowing padding. |
| 2822 | */ |
| 2823 | west = halfwide + STDPAD; |
| 2824 | east = west + parenwidth; |
| 2825 | |
| 2826 | /* |
| 2827 | * Keep trying positions for this paren, working left to |
| 2828 | * right. When we find one that doesn't overlap an existing |
| 2829 | * rectangle, break. This has to succeed at some point, |
| 2830 | * at the rightmost rectangle position if not earlier. |
| 2831 | */ |
| 2832 | for (;;) { |
| 2833 | overlap = NO; |
| 2834 | for (j = 0; j < reclim; j++) { |
| 2835 | /* ignore ones too far north or south */ |
| 2836 | if (rectab[j].relevant == NO) |
| 2837 | continue; |
| 2838 | |
| 2839 | /* if all west or east, okay; else overlap */ |
| 2840 | if (rectab[j].w + FUDGE <= east && |
| 2841 | rectab[j].e >= west + FUDGE) { |
| 2842 | overlap = YES; |
| 2843 | break; |
| 2844 | } |
| 2845 | } |
| 2846 | |
| 2847 | /* if no rectangle overlapped, we found a valid place*/ |
| 2848 | if (overlap == NO) |
| 2849 | break; |
| 2850 | |
| 2851 | /* |
| 2852 | * Something overlapped, so we have to try again. |
| 2853 | * Find the westermost relevant east rectangle boundary |
| 2854 | * that hasn't been tried already, to use as the next |
| 2855 | * trial position for our paren's west. |
| 2856 | */ |
| 2857 | try = -1; |
| 2858 | for (j = 0; j < reclim; j++) { |
| 2859 | /* ignore ones too far north or south */ |
| 2860 | if (rectab[j].relevant == NO || |
| 2861 | rectab[j].tried == YES) |
| 2862 | continue; |
| 2863 | |
| 2864 | /* |
| 2865 | * If this is the first relevant one we haven't |
| 2866 | * tried, or if this is farther west than the |
| 2867 | * westernmost so far, save it as being the |
| 2868 | * new westernmost so far. |
| 2869 | */ |
| 2870 | if (try == -1 || rectab[j].e < rectab[try].e) |
| 2871 | try = j; |
| 2872 | } |
| 2873 | |
| 2874 | if (try == -1) |
| 2875 | pfatal("bug in noterparen()"); |
| 2876 | |
| 2877 | /* |
| 2878 | * Mark this one as having been tried (for next time |
| 2879 | * around, if necessary). Set new trial values for |
| 2880 | * east and west of our paren. |
| 2881 | */ |
| 2882 | rectab[try].tried = YES; |
| 2883 | west = rectab[try].e; |
| 2884 | east = west + parenwidth; |
| 2885 | |
| 2886 | } /* end of while loop trying positions for this acc */ |
| 2887 | |
| 2888 | /* |
| 2889 | * We have the final position for the new paren. Enter it into |
| 2890 | * rectab. Store its east in erparen in the NOTE structure for |
| 2891 | * whichever groups have this note. |
| 2892 | */ |
| 2893 | rectab[reclim].n = north; |
| 2894 | rectab[reclim].s = south; |
| 2895 | rectab[reclim].e = east; |
| 2896 | rectab[reclim].w = west; |
| 2897 | reclim++; |
| 2898 | if (noteptrs[k].top_p != 0) { |
| 2899 | noteptrs[k].top_p->erparen = east; |
| 2900 | } |
| 2901 | if (noteptrs[k].bot_p != 0) { |
| 2902 | noteptrs[k].bot_p->erparen = east; |
| 2903 | } |
| 2904 | |
| 2905 | } /* end of loop for each accidental */ |
| 2906 | |
| 2907 | /* |
| 2908 | * Finally, if the stems were collinear, we have to adjust erparen for |
| 2909 | * all the notes of the bottom group, so that it's relative to the |
| 2910 | * bottom group instead of the top group. |
| 2911 | */ |
| 2912 | if (collinear) { |
| 2913 | for (k = 0; (note_p = GETPTR(k)) != 0; k++) { |
| 2914 | if (noteptrs[k].bot_p != 0) { |
| 2915 | noteptrs[k].bot_p->erparen -= 2 * halfwide |
| 2916 | - W_NORMAL * POINT; |
| 2917 | } |
| 2918 | } |
| 2919 | } |
| 2920 | } |