Cope with commas in numerical fields from the scraper; deal properly with qty in...

[ypp-sc-tools.db-test.git] / yarrg / Commods.pm

diff --git a/yarrg/Commods.pm b/yarrg/Commods.pm
index 44cc74e8543b91b885ee76082cb3ea0042d5f3fd..f74571ddd39c6c5093902c51aae03969cb780532 100644 (file)
--- a/yarrg/Commods.pm
+++ b/yarrg/Commods.pm
@@ -24,6 +24,7 @@ package Commods;
 use IO::File;
 use HTTP::Request::Common ();
 use POSIX;
+use LWP::UserAgent;
 
 use strict;
 use warnings;
@@ -41,7 +42,8 @@ BEGIN {
                      &get_our_version &check_tsv_line
                      &pipethrough_prep &pipethrough_run
                      &pipethrough_run_along &pipethrough_run_finish
-                     &pipethrough_run_gzip
+                     &pipethrough_run_gzip &http_useragent &version_core
+                     &http_useragent_string_map
                      &cgipostform &yarrgpostform &cgi_get_caller
                      &set_ctype_utf8 $masterinfoversion);
     %EXPORT_TAGS = ( );
@@ -49,7 +51,33 @@ BEGIN {
     @EXPORT_OK   = qw();
 }
 
-our $masterinfoversion= 2; # version we understand
+our $masterinfoversion= 2; # version we understand.
+#
+# To extend the source-info.txt format:
+#
+#    * Commods.pm:parse_info1
+#       add code to parse new version
+#
+#    * source-info.txt
+#       add new information
+#
+# If new data should NOT be in master-info.txt too:
+#
+#    * update-master-info:process_some_info
+#       check that code for converting source-info to master-info
+#       removes the extra info; add code to remove it if necessary
+#
+#    * db-idempotent-populate
+#       if database schema is extended, add code to copy data
+#
+# If new data DOES need to be in master-info.txt too:
+#
+#    * Commods.pm:$masterinfoversion
+#       increment
+#
+#    * update-master-info:process_some_info
+#       add code to convert new version to old, by removing
+#       extra info conditionally depending on version
 
 our %oceans; # eg $oceans{'Midnight'}{'Ruby'}{'Eta Island'}= $sources;
 our %clients; # eg $clients{'ypp-sc-tools'}= [ qw(last-page) ];
@@ -74,9 +102,13 @@ my (@rawcm, @nocm); # eg $rawcm[0]='fine rum'; $rawcm[1]='fine %c cloth'
 #  when extending the format of source-info in a non-backward
 #  compatible way, be sure to update update-master-info too.
 
-sub parse_info1 ($$) {
-    my ($mmfn,$src)= @_;
-    my $mm= new IO::File $mmfn, 'r' or die "$mmfn $!";
+sub parse_info1 ($$$) {
+    my ($mmfn,$src,$enoentok)= @_;
+    my $mm= new IO::File $mmfn, 'r';
+    if (!$mm) {
+       return if $enoentok && $!==&ENOENT;
+       die "$mmfn $!";
+    }
     my @ctx= ();
     while (<$mm>) {
        next if m/^\s*\#/;
@@ -176,16 +208,9 @@ sub parse_info1 ($$) {
 }
 
 sub parse_info_clientside () {
-    my $yarrg= $ENV{'YPPSC_YARRG_DICT_UPDATE'};
-    return unless $yarrg;
-    my $master= fetch_with_rsync("info-$masterinfoversion");
-    parse_info1($master,'s');
-    my $local= '_local-info.txt';
-    if (stat $local) {
-       parse_info1($local,'s');
-    } else {
-       die "$local $!" unless $! == &ENOENT;
-    }
+    my $master= fetch_with_rsync("info-v$masterinfoversion");
+    parse_info1($master,'s',1);
+    parse_info1('_local-info.txt','s',1);
 }
 
 sub fetch_with_rsync ($) {
@@ -205,12 +230,12 @@ sub fetch_with_rsync ($) {
 }
 
 sub parse_info_serverside () {
-    parse_info1('source-info.txt','s');
+    parse_info1('source-info.txt','s',0);
 }
 sub parse_info_serverside_ocean ($) {
     my ($oceanname) = @_;
     die "unknown ocean $oceanname ?" unless exists $oceans{$oceanname};
-    parse_info1("_ocean-".(lc $oceanname).".txt",'s');
+    parse_info1("_ocean-".(lc $oceanname).".txt", 's',0);
 }
 
 sub parse_pctb_commodmap () {
@@ -235,16 +260,26 @@ sub get_our_version ($$) {
     my ($aref,$prefix) = @_;
     $aref->{"${prefix}name"}= 'ypp-sc-tools yarrg';
     $aref->{"${prefix}fixes"}= 'lastpage checkpager';
-
-    my $version= `git-describe --tags HEAD || echo 0unknown`; $? and die $?;
-    chomp($version);
-    $aref->{"${prefix}version"}= $version;
+    $aref->{"${prefix}version"}= version_core();
     return $aref;
     # clientname       "ypp-sc-tools"
     # clientversion    2.1-g2e06a26  [from git-describe --tags HEAD]
     # clientfixes      "lastpage"  [space separated list]
 }
 
+sub version_core () {
+    my $version= `
+       if type -p git-describe >/dev/null 2>&1; then
+               gd=git-describe
+       else
+               gd="git describe"
+       fi
+       \$gd --tags HEAD || echo 0unknown
+    `; $? and die $?;
+    chomp($version);
+    return $version;
+}
+
 sub pipethrough_prep () {
     my $tf= IO::File::new_tmpfile() or die $!;
     return $tf;
@@ -382,4 +417,22 @@ sub set_ctype_utf8 () {
     setlocale(LC_CTYPE, "en.UTF-8");
 }
 
+sub http_useragent_string_map ($$) {
+    my ($caller_lib_agent, $reason_style_or_caller) = @_;
+    $caller_lib_agent =~ y/A-Za-z/N-ZA-Mn-za-m/;
+    $caller_lib_agent =~ s/\s/_/g;
+    my $version= version_core();
+    return "yarrg/$version ($reason_style_or_caller)".
+          " $caller_lib_agent".
+          " (http://yarrg.chiark.net/intro)";
+}
+
+sub http_useragent ($) {
+    my ($who) = @_;
+    my $ua= LWP::UserAgent->new;
+    my $base= $ua->_agent();
+    $ua->agent(http_useragent_string_map($base, $who));
+    return $ua;
+}
+
 1;