Improved selector parsing.

Mostly: treating each selection clause seperately instead of having special cases for each el + attribute combination

diff --git a/lib/HTML/Zoom/SelectorParser.pm b/lib/HTML/Zoom/SelectorParser.pm
index 9f72b0a..4e4f11c 100644 (file)
--- a/lib/HTML/Zoom/SelectorParser.pm
+++ b/lib/HTML/Zoom/SelectorParser.pm
@@ -7,26 +7,10 @@ use Carp qw(confess);
 
 my $sel_char = '-\w_';
 my $sel_re = qr/([$sel_char]+)/;
+my $match_value_re = qr/"?$sel_re"?/;
 
-sub new { bless({}, shift) }
-
-my $match_attr_on_regex = sub {
-   my ($self, $name, $attr, $regex) = @_;
-
-   sub {
-      $_[0]->{name} && $_[0]->{name} eq $name and
-      $_[0]->{attrs}{$attr} && $_[0]->{attrs}{$attr} =~ $regex
-   }
-};
-
-my $match_attr_on_eq = sub {
-   my ($self, $name, $attr, $val) = @_;
 
-   sub {
-      $_[0]->{name} && $_[0]->{name} eq $name and
-      $_[0]->{attrs}{$attr} && $_[0]->{attrs}{$attr} eq $val
-   }
-};
+sub new { bless({}, shift) }
 
 sub _raw_parse_simple_selector {
   for ($_[1]) { # same pos() as outside
@@ -36,65 +20,6 @@ sub _raw_parse_simple_selector {
     /\G\*/gc and
       return sub { 1 };
 
-     # 'el[attr~="foo"]
-
-    /\G$sel_re\[$sel_re~="$sel_re"\]/gc and
-      return do {
-        my $name = $1;
-        my $attr = $2;
-        my $val = $3;
-        sub {
-          if (
-            $_[0]->{name} && $_[0]->{name} eq $name and
-            $_[0]->{attrs}{$attr}
-          ) {
-            my %vals = map { $_ => 1 } split /\s+/, $_[0]->{attrs}{$attr};
-            return $vals{$val}
-          }
-          return undef
-        }
-      };
-
-     # 'el[attr^="foo"]
-
-    /\G$sel_re\[$sel_re\^="$sel_re"\]/gc and
-      return do { $_[0]->$match_attr_on_regex($1, $2, qr/^\Q$3\E/) };
-
-     # 'el[attr$="foo"]
-
-    /\G$sel_re\[$sel_re\$="$sel_re"\]/gc and
-      return do { $_[0]->$match_attr_on_regex($1, $2, qr/\Q$3\E$/) };
-
-     # 'el[attr*="foo"]
-
-    /\G$sel_re\[$sel_re\*="$sel_re"\]/gc and
-      return do { $_[0]->$match_attr_on_regex($1, $2, qr/\Q$3\E/) };
-
-     # 'el[attr="foo"]
-
-    /\G$sel_re\[$sel_re="$sel_re"\]/gc and
-      return do { $_[0]->$match_attr_on_eq($1, $2, $3) };
-
-     # 'el[attr]
-
-    /\G$sel_re\[$sel_re\]/gc and
-      return do {
-        my $name = $1;
-        my $attr = $2;
-        sub {
-           $_[0]->{name} && $_[0]->{name} eq $name && $_[0]->{attrs}{$attr}
-        }
-      };
-
-    # 'el.class1' - element + class
-
-    /\G$sel_re\.$sel_re/gc and
-      return do { $_[0]->$match_attr_on_eq($1, 'class', $2) };
-
-    # 'el#id' - element + id
-    /\G$sel_re#$sel_re/gc and
-      return do { $_[0]->$match_attr_on_eq($1, 'id', $2) };
-
     # 'element' - match on tag name
 
     /\G$sel_re/gc and
@@ -123,7 +48,50 @@ sub _raw_parse_simple_selector {
         }
       };
 
-    confess "Couldn't parse $_ as starting with simple selector";
+    # '[attr^=foo]' - match attribute with ^ anchored regex
+    /\G\[$sel_re\^=$match_value_re\]/gc and
+      return do{
+        my $attribute = $1;
+        my $value = $2;
+        $_[0]->{attrs}{$attribute}
+        && $_[0]->{attrs}{$attribute} =~ qr/^\Q$value\E/;
+      };
+
+    # '[attr$=foo]' - match attribute with $ anchored regex
+    /\G\[$sel_re\$=$match_value_re\]/gc and
+      return do{
+        my $attribute = $1;
+        my $value = $2;
+        $_[0]->{attrs}{$attribute}
+        && $_[0]->{attrs}{$attribute} =~ qr/\Q$value\E$/;
+      };
+
+    # '[attr*=foo] - match attribute with regex:
+    /\G\[$sel_re\*=$match_value_re\]/gc and
+      return do{
+        my $attribute = $1;
+        my $value = $2;
+        $_[0]->{attrs}{$attribute}
+        && $_[0]->{attrs}{$attribute} =~ qr/\Q$value\E$/;
+      };
+
+    # '[attr=bar]' - match attributes
+    /\G\[$sel_re=$match_value_re\]/gc and
+      return do {
+        my $attribute = $1;
+        my $value = $2;
+        sub{
+          $_[0]->{attrs}{$attribute}
+          && $_[0]->{attrs}{$attribute} eq $value;
+        }
+      };
+
+    # '[attr] - match attribute being present:
+    /\G\[$sel_re\]/gc and
+      return do {
+        my $attribute = $1;
+        $_[0]->{attrs}{$attribute};
+      }
   }
 }
 
@@ -135,10 +103,40 @@ sub parse_selector {
   for ($sel) {
     my @sub;
     PARSE: { do {
-      push(@sub, $self->_raw_parse_simple_selector($_));
-      last PARSE if (pos == length);
-      /\G\s*,\s*/gc or confess "Selectors not comma separated";
-    } until (pos == length) };
+
+      my @this_chain;
+
+      # slurp selectors until we find something else:
+      while( my $sel = $self->_raw_parse_simple_selector($_) ){
+        push @this_chain, $sel;
+      }
+
+      if( @this_chain == 1 )
+      {
+        push @sub, @this_chain;
+      }
+      else{
+        # make a compound match closure of everything
+        # in this chain of selectors:
+        push @sub, sub{
+          my $r;
+          for my $inner ( @this_chain ){
+            if( ! ($r = $inner->( @_ )) ){
+              return $r;
+            }
+          }
+          return $r;
+        }
+      }
+
+      # now we're at the end or a delimiter:
+      last PARSE if( pos == length );
+      /\G\s*,\s*/gc or do {
+        /\G(.*)/;
+        $self->_blam( "Selectors not comma separated." );
+      }
+
+     } until (pos == length) };
     return $sub[0] if (@sub == 1);
     return sub {
       foreach my $inner (@sub) {
@@ -149,4 +147,12 @@ sub parse_selector {
 }
 
 
+sub _blam {
+  my ($self, $error) = @_;
+  my $hat = (' ' x (pos||0)).'^';
+  die "Error parsing dispatch specification: ${error}\n
+${_}
+${hat} here\n";
+}
+
 1;

diff --git a/t/selectors.t b/t/selectors.t
index 01e11cf..59d9c7c 100644 (file)
--- a/t/selectors.t
+++ b/t/selectors.t
@@ -4,6 +4,16 @@ use Test::More;
 
 use HTML::Zoom;
 
+my $tmpl = <<END;
+<body>
+  <div class="main">
+    <span prop='moo' class="hilight name">Bob</span>
+    <span class="career">Builder</span>
+    <hr />
+  </div>
+</body>
+END
+
 # el#id
 is( HTML::Zoom->from_html('<div id="yo"></div>')
    ->select('div#yo')
@@ -36,6 +46,15 @@ is( HTML::Zoom->from_html('<div frew="yo"></div>')
    '<div frew="yo">grg</div>',
    'E[attr="val"] works' );
 
+# el[attr=foo]
+is( HTML::Zoom->from_html('<div frew="yo"></div>')
+    ->select('div[frew=yo]')
+    ->replace_content('grg')
+    ->to_html,
+    '<div frew="yo">grg</div>',
+    'E[attr=val] works' );
+ 
+
 # el[attr*="foo"]
 is( HTML::Zoom->from_html('<div f="frew goog"></div>')
    ->select('div[f*="oo"]')
@@ -68,6 +87,14 @@ is( HTML::Zoom->from_html('<div f="foo bar"></div>')
    '<div f="foo bar">grg</div>',
    'E[attr*="val"] works' );
 
+# [attr=bar]
+ok( check_select( '[prop=moo]'), '[attr=bar]' );
+
+# el[attr=bar],[prop=foo]
+is( check_select('span[class=career],[prop=moo]'), 2,
+    'Multiple selectors: el[attr=bar],[attr=foo]');
+
+
 # sel1 sel2
 is( HTML::Zoom->from_html('<table><tr></tr><tr></tr></table>')
    ->select('table tr')
@@ -86,3 +113,16 @@ is( HTML::Zoom->from_html('<table><tr><td></td></tr><tr><td></td></tr></table>')
    'sel1 sel2 sel3 works' );
 
 done_testing;
+
+
+sub check_select{
+    # less crude?:
+    my $output = HTML::Zoom
+    ->from_html($tmpl)
+    ->select(shift)->replace("the monkey")->to_html;
+    my $count = 0;
+    while ( $output =~ /\G?.*the monkey/gc ){
+        $count++;
+    }
+    return $count;
+}