Added a generate_snippet mode.

Joshua Pokotilow · Joshua Pokotilow · commit cb0926aa382b · 2011-12-09T17:29:34.000-08:00
When passing the --generate-snippet parameter to the
program, only the contents of the HTML &lt;body&gt; tag are
printed. This may be useful for situations when markup
needs to be formatted to work with AJAX services, or for
other situations where it's inappropriate to print a
whole document.
diff --git a/src/html2xhtml.c b/src/html2xhtml.c
@@ -211,6 +211,8 @@ static void process_parameters(int argc, char **argv)
     } else if (!strcmp(argv[i], "--lcs")) {
       charset_dump_aliases(stdout);
       exit(0);
+    } else if (!strcmp(argv[i], "--generate-snippet")) {
+      param_generate_snippet = 1;
     } else {
       help();
       exit(1);
diff --git a/src/params.c b/src/params.c
@@ -36,6 +36,7 @@ int   param_compact_block_elms;
 int   param_compact_empty_elm_tags;
 int   param_empty_tags;
 int   param_crlf_eol;
+int   param_generate_snippet;
 
 void params_set_defaults()
 {
@@ -55,5 +56,6 @@ void params_set_defaults()
   param_cgi_html_output = 0;
   param_empty_tags = 0;
   param_crlf_eol = 0;
+  param_generate_snippet = 0;
 }
 
diff --git a/src/params.h b/src/params.h
@@ -47,6 +47,7 @@ extern int   param_compact_block_elms;
 extern int   param_compact_empty_elm_tags;
 extern int   param_empty_tags;
 extern int   param_crlf_eol;
+extern int   param_generate_snippet;
 
 void params_set_defaults(void);
 
diff --git a/src/procesador.c b/src/procesador.c
@@ -2113,17 +2113,18 @@ static void write_document(document_t *doc)
 
   cprintf_init(param_charset_out, param_outputf);
 
-  /* write <?xml... */
-  cprintf("%s?xml version=\"1.0\"", lt);
-/*   if (document->encoding[0])  */
-/*     cprintf(" encoding=\"%s\"",document->encoding); */
-  cprintf(" encoding=\"%s\"", param_charset_out->preferred_name);
-  cprintf("?%s%s%s", gt, eol, eol);
-
-  /* write <!DOCTYPE... */
-  cprintf("%s!DOCTYPE html%s   %s%s   \"%s\" %s%s",
-	  lt, eol, doctype_string[doctype], eol, dtd_string[doctype], gt, eol);
-  
+  if (!param_generate_snippet) {
+      /* write <?xml... */
+      cprintf("%s?xml version=\"1.0\"", lt);
+    /*   if (document->encoding[0])  */
+    /*     cprintf(" encoding=\"%s\"",document->encoding); */
+      cprintf(" encoding=\"%s\"", param_charset_out->preferred_name);
+      cprintf("?%s%s%s", gt, eol, eol);
+
+      /* write <!DOCTYPE... */
+      cprintf("%s!DOCTYPE html%s   %s%s   \"%s\" %s%s",
+          lt, eol, doctype_string[doctype], eol, dtd_string[doctype], gt, eol);
+  }
   p = doc->inicio;
   indent = 0;
   write_node(p);
@@ -2157,14 +2158,29 @@ static int write_node(tree_node_t *node)
   return len;
 }
 
+static int should_write_element(tree_node_t* nodo)
+{
+    return !(param_generate_snippet && (ELM_ID(nodo) == ELMID_HEAD));
+}
+
+static int should_write_tags(tree_node_t* nodo)
+{
+    return !(param_generate_snippet && (ELM_ID(nodo) == ELMID_HTML
+                                        || ELM_ID(nodo) == ELMID_BODY));
+}
 
 static int write_element(tree_node_t *elm)
 {
   int len = 0;
   tree_node_t *n;
   int is_block;
   int xml_space_activated;
+  int writes_tags;
 
+  if (!should_write_element(elm))
+    return 0;
+    
+  writes_tags = should_write_tags(elm);
   is_block = dtd_elm_is_block(ELM_ID(elm));
 
   /* activate "xml:space preserve" if necessary */
@@ -2183,45 +2199,50 @@ static int write_element(tree_node_t *elm)
   }
 
   /* write start tag */
-  if (is_block) {
-     len += write_indent(indent, 1);
-     inline_on = 0;
-  } else {
-    if (!inline_on) {
-      inline_on = 1;
-      whitespace_needed = 0;
-      if (!param_compact_block_elms)
-	len += write_indent(indent, 1);
+//  if (param_generate_snippet && elm == html
+  if (writes_tags) {
+    if (is_block) {
+       len += write_indent(indent, 1);
+       inline_on = 0;
+    } else {
+      if (!inline_on) {
+        inline_on = 1;
+        whitespace_needed = 0;
+        if (!param_compact_block_elms)
+      len += write_indent(indent, 1);
+      }
     }
+    len += write_start_tag(elm);
   }
-  len += write_start_tag(elm);
 
   /* process children nodes */
   n = elm->cont.elemento.hijo;
-  if (is_block)
+  if (is_block && writes_tags)
     indent += param_tab_len;
   while (n) {
     len += write_node(n);
     n = n->sig;
   }
-  if (is_block)
+  if (is_block && writes_tags)
     indent -= param_tab_len;
 
   /* write end tag if not empty */
-  if (elm->cont.elemento.hijo) {
+  if (writes_tags) {
+    if (elm->cont.elemento.hijo) {
     if (is_block) {
       if (inline_on) {
-	inline_on = 0;
-	if (!param_compact_block_elms)
-	  len += write_indent(indent, 1);
+    inline_on = 0;
+    if (!param_compact_block_elms)
+      len += write_indent(indent, 1);
       } else {
-	len += write_indent(indent, 1);
+    len += write_indent(indent, 1);
       }
     }
     len += write_end_tag(elm);
-  } else if (!param_empty_tags
-	     && elm_list[ELM_ID(elm)].contenttype[doctype] != CONTTYPE_EMPTY) {
+    } else if (!param_empty_tags
+         && elm_list[ELM_ID(elm)].contenttype[doctype] != CONTTYPE_EMPTY) {
     len += write_end_tag(elm);
+    }
   }
 
   /* deactivate "xml:space preserve" if activated */
@@ -2403,7 +2424,6 @@ static int write_start_tag(tree_node_t* nodo)
   char *elm_name;
   int elm_name_len;
   int printed;
-  
 
   elm_name = elm_list[ELM_ID(nodo)].name;
   elm_name_len = strlen(elm_name);

Original file line number	Diff line number	Diff line change
`@@ -36,6 +36,7 @@ int param_compact_block_elms;`
`36`	`36`	`int param_compact_empty_elm_tags;`
`37`	`37`	`int param_empty_tags;`
`38`	`38`	`int param_crlf_eol;`
	`39`	`+int param_generate_snippet;`
`39`	`40`
`40`	`41`	`void params_set_defaults()`
`41`	`42`	`{`
`@@ -55,5 +56,6 @@ void params_set_defaults()`
`55`	`56`	`param_cgi_html_output = 0;`
`56`	`57`	`param_empty_tags = 0;`
`57`	`58`	`param_crlf_eol = 0;`
	`59`	`+ param_generate_snippet = 0;`
`58`	`60`	`}`
`59`	`61`