From: Vladimír Vondruš <mosra@centrum.cz>
Date: Thu, 18 Jul 2019 15:24:50 +0000 (+0200)
Subject: documentation: trim spaces from the right only if nothing is found.
X-Git-Url: https://www.chiark.greenend.org.uk/ucgi/~cjwatson/git?a=commitdiff_plain;h=20e1743f2dfa1696fafd0907f44385e0266845a5;p=blog.git

documentation: trim spaces from the right only if nothing is found.

With unconditional trimming it was not possible to narrow down results
for page titles containing spaces (or subpages, which are separated by
double right arrow surrounded by spaces).
---

diff --git a/documentation/search.js b/documentation/search.js
index 8b8d7828..67e8e703 100644
--- a/documentation/search.js
+++ b/documentation/search.js
@@ -232,8 +232,10 @@ var Search = {
     /* Returns the values in UTF-8, but input is in whatever shitty 16bit
        encoding JS has */
     search: function(searchString) {
-        /* Normalize the search string first, convert to UTF-8 */
-        searchString = this.toUtf8(searchString.toLowerCase().trim());
+        /* Normalize the search string first, convert to UTF-8 and trim spaces
+           from the left. From the right they're trimmed only if nothing is
+           found, see below. */
+        searchString = this.toUtf8(searchString.toLowerCase().replace(/^\s+/,''));
 
         /* TODO: maybe i could make use of InputEvent.data and others here */
 
@@ -277,8 +279,18 @@ var Search = {
                 break;
             }
 
-            /* Character not found, exit */
-            if(!found) break;
+            /* Character not found */
+            if(!found) {
+                /* If we found everything except spaces at the end, pretend the
+                   spaces aren't there. On the other hand, we *do* want to
+                   try searching with the spaces first -- it can narrow down
+                   the result list for page names or show subpages (which are
+                   after a lookahead barrier that's a space). */
+                if(!searchString.substr(foundPrefix).trim().length)
+                    searchString = searchString.substr(0, foundPrefix);
+
+                break;
+            }
         }
 
         /* Save the whole found prefix for next time */
diff --git a/documentation/test/js-test-data/searchdata.b85 b/documentation/test/js-test-data/searchdata.b85
index 1de2e518..1a915d15 100644
--- a/documentation/test/js-test-data/searchdata.b85
+++ b/documentation/test/js-test-data/searchdata.b85
@@ -1 +1 @@
-O+!-w2LOZt004pl003kG000310RR921ONaj009U904M+f4gdgd009&L0BHdL0{{R4AOHX<00ATb04M+fDgXd(00A%n0BHaLHUI!^00BGz06GBy0suk)fI0vHNB{tG00B?{0B-;RRsaBW00CS80Am0FVgLYT0RRO600C|Q04V?gasU7*00DRa0B!&QegFVz00D#m0BryPiU0sQ0RaR6kN|)>00EW&0A&CHo&W%600E=`0B!&QssI3C00SBT0BvXh0Cund0CE5Uwg3P+0RaF2!~lRg00GJX0B8UK(f|N-0{{U40{{g800G_r04V?g<^TXF00Ha(0B!&R*Z=@w@&Ev70RRU8009C40A&CH1_1zU009gE0A~OJ5&-~i0RadA7y$rb00ABW0CWHWCIJ9r00OE20AVZv0A&FH1^@s7JOKb@00BS&0A~OJMgag}00B$^0B`^SQUL&B00CG50CfNa_y7QHXaE3qG64W`UI74eC;$K;KL7wpR{#JydjJ4QkpKWVvj6~1$p8Ql<p2Nx@Bjb_{r~^~O<{Cs0B&JzWpi+0V`WWYbZ9PUbZu+^01^l~I&EogC~0nVEFfuabSVHMZE0=*0025VR%K&!Z*l-*Y+-YAO<{CsUol@XR%K&!Z*neZbZu+`02l^3I&EogC@COgZ*FsR03&T_ZU6uPIyzEeZf9ixV{Bn_b4_7%XkRg3F;Zb}XJsyEbZu+|02c;2I&EogC@COgZ*FsR03&T_Zct%oWgx=4AX9Z>aA9X<0CRO>aA9X<E@*UZYy<#OWn*+<Zf9&|1ONvB00KHXQe|UwC@BI80S*Bd1snh%aA9X<ZeeX@b8ul}Wn*k%b8}{OZesud
\ No newline at end of file
+O+!-w2LQSO007AX005Q&000310RR921ONaj009U904M+f4gdgd009&L0BHdL0{{R4AOHX<00ATb04M+fDgXd(00A%n0BHaLHUI!^00BGz06GBy0suk)fI0vHNB{tG00B?{0B-;RRsaBW00CS80Am0FVgLYT0RRO600C|Q04V?gasU7*00DRa0B!&QegFVz00D#m0BryPiU0sQ0RaR6kN|)>00EW&0A&CHo&W%600E=`0B!&QssI3C00SBT0BvXh0Cund0CE5Uwg3P+0RaF2!~lRg00GJX0B8UK(f|N-0{{U40{{g800G_r04V?g<^TXF00Ha(0B!&R*Z=@w@&Ev70RRX9009C40A&CH1_1zU009gE0A~OJ5&-~i0RagB7y$rb00ABW0CWHWCIJ9r00OE20AVZv0A&FH2LJ#8JOKb@00BS&0A~OJMgag}00B$^0B`^SQUL&B00CG50CfNXUI74e00CqH03ZMXY5@Sd00D3T0Kx$Q1^{*efFJ+?d;tJu00D#n0A~OJiU9y&00sB}0BvXh0Cq9~0CJE40B~Lb0COw=03bsE07+W_06KpF07;bq064b*08PyR01(>%02uZF0003200|EP002#4bZ7u>VQpn|aA9L*O<{CsE@*UZYybcf2s%1#X>KTKZgealX>N2W03&T_ZU6uPIyzQmV{~tF0Ap-nb8}5$bZB2OUolo?V{~tFE@*UZYyton20A)zX>KSfAY*TCb94YBZE0=*0025VQekdqWdLJrVRLg$VRUF;F<&uKVQyz-E@*UZYy<!o20A)zX>KSfAY*TCb94YBZE0>$VP|CkaA9X<E@*UZYz6=TAi}#KQ*~l+VP|Ckb9G{HVP|D7Xmo9C1OQTHV{~C|XKZBz00;m80y;WUWn*+GDFO-s4gnVh8~`A2VP|D-VQpn|aA9L*V{Bn_b7pmJV*mgE
\ No newline at end of file
diff --git a/documentation/test/js-test-data/searchdata.bin b/documentation/test/js-test-data/searchdata.bin
index dfc16df1..bf52b592 100644
Binary files a/documentation/test/js-test-data/searchdata.bin and b/documentation/test/js-test-data/searchdata.bin differ
diff --git a/documentation/test/populate-js-test-data.py b/documentation/test/populate-js-test-data.py
index 9903ee84..a6410a94 100755
--- a/documentation/test/populate-js-test-data.py
+++ b/documentation/test/populate-js-test-data.py
@@ -70,7 +70,10 @@ index = map.add("Math::Range::min() const", "classMath_1_1Range.html#min", suffi
 trie.insert("math::range::min()", index, lookahead_barriers=[4, 11])
 trie.insert("range::min()", index, lookahead_barriers=[5])
 trie.insert("min()", index)
-trie.insert("subpage", map.add("Page » Subpage", "subpage.html", flags=ResultFlag.from_type(ResultFlag.NONE, EntryType.PAGE)))
+trie.insert("page", map.add("Page", "page.html", flags=ResultFlag.from_type(ResultFlag.NONE, EntryType.PAGE)))
+index = map.add("Page » Subpage", "subpage.html", flags=ResultFlag.from_type(ResultFlag.NONE, EntryType.PAGE))
+trie.insert("page » subpage", index, lookahead_barriers=[4])
+trie.insert("subpage", index)
 
 trie.insert("rectangle", map.add("Rectangle", "", alias=range_index))
 trie.insert("rect", map.add("Rectangle::Rect()", "", suffix_length=2, alias=range_index))
diff --git a/documentation/test/test-search.js b/documentation/test/test-search.js
index ec45d218..0efdd41d 100644
--- a/documentation/test/test-search.js
+++ b/documentation/test/test-search.js
@@ -79,7 +79,7 @@ const { StringDecoder } = require('string_decoder');
 /* Verify that base85-decoded file is equivalent to the binary */
 {
     let binary = fs.readFileSync(path.join(__dirname, "js-test-data/searchdata.bin"));
-    assert.equal(binary.byteLength, 674);
+    assert.equal(binary.byteLength, 745);
     let b85 = fs.readFileSync(path.join(__dirname, "js-test-data/searchdata.b85"), {encoding: 'utf-8'});
     assert.deepEqual(new DataView(binary.buffer.slice(binary.byteOffset, binary.byteOffset + binary.byteLength)), new DataView(Search.base85decode(b85), 0, binary.byteLength));
 }
@@ -115,7 +115,7 @@ const { StringDecoder } = require('string_decoder');
 {
     let buffer = fs.readFileSync(path.join(__dirname, "js-test-data/searchdata.bin"));
     assert.ok(Search.init(buffer.buffer.slice(buffer.byteOffset, buffer.byteOffset + buffer.byteLength)));
-    assert.equal(Search.dataSize, 674);
+    assert.equal(Search.dataSize, 745);
     assert.equal(Search.symbolCount, "7 symbols (0.7 kB)");
     assert.equal(Search.maxResults, 100);
 
@@ -181,9 +181,6 @@ const { StringDecoder } = require('string_decoder');
           suffixLength: 3 }], 'tor'];
     assert.deepEqual(Search.search('vec'), resultsForVec);
 
-    /* Uppercase things and spaces */
-    assert.deepEqual(Search.search(' Vec  '), resultsForVec);
-
     /* Not found */
     assert.deepEqual(Search.search('pizza'), [[], '']);
 
@@ -191,7 +188,7 @@ const { StringDecoder } = require('string_decoder');
     assert.deepEqual(Search.search('su'), [[
         { name: Search.toUtf8('Page » Subpage'),
           url: 'subpage.html',
-          flags: 0,
+          flags: 8, /* has prefix */
           cssClass: 'm-success',
           typeName: 'page',
           suffixLength: 5 }], 'bpage']);
@@ -220,11 +217,63 @@ const { StringDecoder } = require('string_decoder');
           suffixLength: 8 }], '']);
 }
 
+/* Search with spaces */
+{
+    let buffer = fs.readFileSync(path.join(__dirname, "js-test-data/searchdata.bin"));
+    assert.ok(Search.init(buffer.buffer.slice(buffer.byteOffset, buffer.byteOffset + buffer.byteLength)));
+    assert.equal(Search.dataSize, 745);
+    assert.equal(Search.symbolCount, "7 symbols (0.7 kB)");
+    assert.equal(Search.maxResults, 100);
+
+    /* No spaces */
+    assert.deepEqual(Search.search('pa'), [[
+        { name: Search.toUtf8('Page'),
+          url: 'page.html',
+          flags: 0,
+          cssClass: 'm-success',
+          typeName: 'page',
+          suffixLength: 2 }], 'ge']);
+
+    /* Spaces from the left are always trimmed */
+    assert.deepEqual(Search.search('  \t pa'), [[
+        { name: Search.toUtf8('Page'),
+          url: 'page.html',
+          flags: 0,
+          cssClass: 'm-success',
+          typeName: 'page',
+          suffixLength: 2 }], 'ge']);
+
+    /* Spaces from the right are trimmed if nothing is found */
+    assert.deepEqual(Search.search('pa \n '), [[
+        { name: Search.toUtf8('Page'),
+          url: 'page.html',
+          flags: 0,
+          cssClass: 'm-success',
+          typeName: 'page',
+          suffixLength: 2 }], 'ge']);
+
+    /* But not if they have results */
+    assert.deepEqual(Search.search('page'), [[
+        { name: Search.toUtf8('Page'),
+          url: 'page.html',
+          flags: 0,
+          cssClass: 'm-success',
+          typeName: 'page',
+          suffixLength: 0 }], '']);
+    assert.deepEqual(Search.search('page '), [[
+        { name: Search.toUtf8('Page » Subpage'),
+          url: 'subpage.html',
+          flags: 8, /* has prefix */
+          cssClass: 'm-success',
+          typeName: 'page',
+          suffixLength: 10 }], Search.toUtf8('» subpage')]);
+}
+
 /* Search, limiting the results to 3 */
 {
     let buffer = fs.readFileSync(path.join(__dirname, "js-test-data/searchdata.bin"));
     assert.ok(Search.init(buffer.buffer.slice(buffer.byteOffset, buffer.byteOffset + buffer.byteLength), 3));
-    assert.equal(Search.dataSize, 674);
+    assert.equal(Search.dataSize, 745);
     assert.equal(Search.symbolCount, "7 symbols (0.7 kB)");
     assert.equal(Search.maxResults, 3);
     assert.deepEqual(Search.search('m'), [[
@@ -252,7 +301,7 @@ const { StringDecoder } = require('string_decoder');
 {
     let b85 = fs.readFileSync(path.join(__dirname, "js-test-data/searchdata.b85"), {encoding: 'utf-8'});
     assert.ok(Search.load(b85));
-    assert.equal(Search.dataSize, 676); /* some padding on the end, that's okay */
+    assert.equal(Search.dataSize, 748); /* some padding on the end, that's okay */
     assert.equal(Search.symbolCount, "7 symbols (0.7 kB)");
     assert.equal(Search.maxResults, 100);
     assert.deepEqual(Search.search('min'), [[