[bug-gettext] [PATCH] its: Add new preserveSpaceRule "paragraph"

[Daiki Ueno]

This implements a new text extraction rule preserving paragraph
boundaries, as mentioned in:
https://gitlab.gnome.org/GNOME/glib/issues/1350

* gettext-tools/doc/gettext.texi (Preparing ITS Rules): Mention "paragraph".
* gettext-tools/src/its.c (its_rule_list_extract_text): Accept "paragraph".
(its_merge_context_merge_node): Likewise.
(normalize_whitespace): Handle "paragraph" rule.
* gettext-tools/src/its.h (ITS_WHITESPACE_NORMALIZE_PARAGRAPH): New enum value.
* gettext-tools/tests/xgettext-its-1: Add test for "paragraph" rule.
---
 gettext-tools/doc/gettext.texi     | 12 +++---
 gettext-tools/src/its.c            | 64 +++++++++++++++++++++++++++++-
 gettext-tools/src/its.h            |  1 +
 gettext-tools/tests/xgettext-its-1 | 50 +++++++++++++++++++++++
 4 files changed, 121 insertions(+), 6 deletions(-)

diff --git a/gettext-tools/doc/gettext.texi b/gettext-tools/doc/gettext.texi
index 892854c75..306605842 100644
--- a/gettext-tools/doc/gettext.texi
+++ b/gettext-tools/doc/gettext.texi
@@ -12354,10 +12354,12 @@ A required @code{escape} attribute with the value 
@code{yes} or @code{no}.
 @item Extended Preserve Space
 
 This data category extends the standard @samp{Preserve Space} data
-category with the additional value @samp{trim}.  The value means to
-remove the leading and trailing whitespaces of the content, but not to
-normalize whitespaces in the middle.  In the global rule, the
address@hidden element contains the following:
+category with the additional values @samp{trim} and @samp{paragraph}.
address@hidden means to remove the leading and trailing whitespaces of the
+content, but not to normalize whitespaces in the middle.
address@hidden means to normalize the content but keep the paragraph
+boundaries.  In the global
+rule, the @code{preserveSpaceRule} element contains the following:
 
 @itemize
 @item
@@ -12366,7 +12368,7 @@ that selects the nodes to which this rule applies.
 
 @item
 A required @code{space} attribute with the value @code{default},
address@hidden, or @code{trim}.
address@hidden, @code{trim}, or @code{paragraph}.
 @end itemize
 
 @end table
diff --git a/gettext-tools/src/its.c b/gettext-tools/src/its.c
index 9b4b397e5..445c0d9b5 100644
--- a/gettext-tools/src/its.c
+++ b/gettext-tools/src/its.c
@@ -399,6 +399,60 @@ normalize_whitespace (const char *text, enum 
its_whitespace_type_ty whitespace)
     case ITS_WHITESPACE_TRIM:
       return trim (text);
 
+    case ITS_WHITESPACE_NORMALIZE_PARAGRAPH:
+      /* Normalize whitespaces within the text, keeping paragraph
+         boundaries.  */
+      {
+        char *result, *p, *end;
+
+        result = xstrdup (text);
+        end = result + strlen (result);
+        for (p = result; *p != '\0';)
+          {
+            char *pp, *pend = NULL;
+            int c;
+
+            /* Find a paragraph boundary.  */
+            for (pp = p; *pp != '\0';)
+              {
+                char *nl = strchrnul (pp, '\n');
+                if (*nl == '\0')
+                  {
+                    pend = nl;
+                    break;
+                  }
+                else if (*(nl + 1) == '\n')
+                  {
+                    pend = nl;
+                    break;
+                  }
+                pp = nl + 1;
+              }
+
+            /* Normalize whitespaces in the paragraph.  */
+            assert (pend != NULL);
+            c = *pend;
+            *pend = '\0';
+            for (pp = p; *pp != '\0';)
+              {
+                size_t len = strspn (pp, " \t\n");
+                if (len > 0)
+                  {
+                    *pp = ' ';
+                    memmove (pp + 1, pp + len, end - (pp + len));
+                    end -= len - 1;
+                    *end = '\0';
+                    pend -= len - 1;
+                    *pend = '\0';
+                    pp++;
+                  }
+                pp += strcspn (pp, " \t\n");
+              }
+            *pend = c;
+            p = pp + strspn (pend, "\n");
+          }
+        return result;
+      }
     default:
       /* Normalize whitespaces within the text, but not at the beginning
          nor the end of the text.  */
@@ -996,7 +1050,11 @@ its_preserve_space_rule_constructor (struct its_rule_ty 
*pop,
            || strcmp (prop, "default") == 0
            /* gettext extension: remove leading/trailing whitespaces only.  */
            || (node->ns && xmlStrEqual (node->ns->href, BAD_CAST GT_NS)
-               && strcmp (prop, "trim") == 0)))
+               && strcmp (prop, "trim") == 0)
+           /* gettext extension: same as default except keeping
+              paragraph boundaries.  */
+           || (node->ns && xmlStrEqual (node->ns->href, BAD_CAST GT_NS)
+               && strcmp (prop, "paragraph") == 0)))
     {
       error (0, 0, _("invalid attribute value \"%s\" for \"%s\""),
              prop, "space");
@@ -1715,6 +1773,8 @@ its_rule_list_extract_text (its_rule_list_ty *rules,
         whitespace = ITS_WHITESPACE_PRESERVE;
       else if (value && strcmp (value, "trim") == 0)
         whitespace = ITS_WHITESPACE_TRIM;
+      else if (value && strcmp (value, "paragraph") == 0)
+        whitespace = ITS_WHITESPACE_NORMALIZE_PARAGRAPH;
       else
         whitespace = ITS_WHITESPACE_NORMALIZE;
 
@@ -1842,6 +1902,8 @@ its_merge_context_merge_node (struct its_merge_context_ty 
*context,
         whitespace = ITS_WHITESPACE_PRESERVE;
       else if (value && strcmp (value, "trim") == 0)
         whitespace = ITS_WHITESPACE_TRIM;
+      else if (value && strcmp (value, "paragraph") == 0)
+        whitespace = ITS_WHITESPACE_NORMALIZE_PARAGRAPH;
       else
         whitespace = ITS_WHITESPACE_NORMALIZE;
 
diff --git a/gettext-tools/src/its.h b/gettext-tools/src/its.h
index 72c30c992..49af5cec5 100644
--- a/gettext-tools/src/its.h
+++ b/gettext-tools/src/its.h
@@ -33,6 +33,7 @@ enum its_whitespace_type_ty
 {
   ITS_WHITESPACE_PRESERVE,
   ITS_WHITESPACE_NORMALIZE,
+  ITS_WHITESPACE_NORMALIZE_PARAGRAPH,
   ITS_WHITESPACE_TRIM
 };
 
diff --git a/gettext-tools/tests/xgettext-its-1 
b/gettext-tools/tests/xgettext-its-1
index 125f3e682..5fe70b366 100755
--- a/gettext-tools/tests/xgettext-its-1
+++ b/gettext-tools/tests/xgettext-its-1
@@ -171,6 +171,29 @@ cat <<\EOF >messages.xml
   <message unescaped="This is an unescaped attribute &lt;&gt;&amp;&quot;">
     <p></p>
   </message>
+  <message>
+    <p xml:space="paragraph">
+    This is the first paragraph with
+a newline.
+
+    This is  the  second paragprah with spaces.
+
+    This is the last paragraph.</p>
+  </message>
+  <message>
+    <p xml:space="paragraph">This is the only one paragraph</p>
+  </message>
+  <message>
+    <p xml:space="paragraph">This is the only one paragraph with a boundary
+
+</p>
+  </message>
+  <message>
+    <p xml:space="paragraph"></p>
+  </message>
+  <message>
+    <p xml:space="paragraph"> </p>
+  </message>
 </messages>
 EOF
 
@@ -247,6 +270,33 @@ msgstr ""
 #: messages.xml:61
 msgid "This is an unescaped attribute <>&\""
 msgstr ""
+
+#. (itstool) path: message/p
+#: messages.xml:65
+msgid ""
+" This is the first paragraph with a newline.\n"
+"\n"
+" This is the second paragprah with spaces.\n"
+"\n"
+" This is the last paragraph."
+msgstr ""
+
+#. (itstool) path: message/p
+#: messages.xml:74
+msgid "This is the only one paragraph"
+msgstr ""
+
+#. (itstool) path: message/p
+#: messages.xml:77
+msgid ""
+"This is the only one paragraph with a boundary\n"
+"\n"
+msgstr ""
+
+#. (itstool) path: message/p
+#: messages.xml:85
+msgid " "
+msgstr ""
 EOF
 
 : ${DIFF=diff}
-- 
2.20.1