Merge with git+ssh://pasky.or.cz/srv/git/elinks.git

src/dom/configuration.c

@@ -1,5 +1,9 @@
 /* DOM Configuration */
 
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
 #include "elinks.h"
 
 #include "dom/configuration.h"

src/dom/node.c

@@ -190,7 +190,8 @@ dom_node_list_bsearch(struct dom_node_search *search, struct dom_node_list *list
 	return NULL;
 }
 
-int get_dom_node_map_index(struct dom_node_list *list, struct dom_node *node)
+int
+get_dom_node_map_index(struct dom_node_list *list, struct dom_node *node)
 {
 	struct dom_node_search search = INIT_DOM_NODE_SEARCH(node, list);
 	struct dom_node *match = dom_node_list_bsearch(&search, list);
@@ -267,6 +268,24 @@ get_dom_node_prev(struct dom_node *node)
 	return NULL;
 }
 
+struct dom_node *
+get_dom_node_next(struct dom_node *node)
+{
+	struct dom_node_list **list;
+	int index;
+
+	assert(node->parent);
+
+	list = get_dom_node_list(node->parent, node);
+	if (!list) return NULL;
+
+	index = get_dom_node_list_pos(*list, node);
+	if (index + 1 < (*list)->size)
+		return (*list)->entries[index + 1];
+
+	return NULL;
+}
+
 struct dom_node *
 get_dom_node_child(struct dom_node *parent, enum dom_node_type type,
 		   int16_t subtype)
@@ -327,6 +346,23 @@ init_dom_node_(unsigned char *file, int line,
 	node->type   = type;
 	node->parent = parent;
 
+	/* Make it possible to add a node to a parent without allocating the
+	 * strings. */
+	if (allocated >= 0) {
+		node->allocated = !!allocated;
+	} else if (parent) {
+		node->allocated = parent->allocated;
+	}
+
+	if (node->allocated) {
+		if (!init_dom_string(&node->string, string->string, string->length)) {
+			done_dom_node(node);
+			return NULL;
+		}
+	} else {
+		copy_dom_string(&node->string, string);
+	}
+
 	if (parent) {
 		struct dom_node_list **list = get_dom_node_list(parent, node);
 		int sort = (type == DOM_NODE_ATTRIBUTE);
@@ -342,22 +378,6 @@ init_dom_node_(unsigned char *file, int line,
 			done_dom_node(node);
 			return NULL;
 		}
-
-		/* Make it possible to add a node to a parent without
-		 * allocating the strings. */
-		node->allocated = allocated < 0 ? parent->allocated : !!allocated;
-
-	} else if (allocated >= 0) {
-			node->allocated = !!allocated;
-	}
-
-	if (node->allocated) {
-		if (!init_dom_string(&node->string, string->string, string->length)) {
-			done_dom_node(node);
-			return NULL;
-		}
-	} else {
-		copy_dom_string(&node->string, string);
 	}
 
 	return node;

src/dom/node.h

@@ -111,6 +111,9 @@ struct dom_attribute_node {
 	 * to reduce string comparing and only do one fast find mapping. */
 	uint16_t type;
 
+	/* The attribute value is delimited by quotes. Can be NUL, ' or ". */
+	unsigned char quoted;
+
 	/* Was the attribute specified in the DTD as a default attribute or was
 	 * it added from the document source. */
 	unsigned int specified:1;
@@ -124,9 +127,6 @@ struct dom_attribute_node {
 
 	/* The attribute value references some other resource */
 	unsigned int reference:1;
-
-	/* The attribute value is delimited by quotes */
-	unsigned int quoted:1;
 };
 
 struct dom_text_node {
@@ -242,6 +242,9 @@ int get_dom_node_map_index(struct dom_node_list *list, struct dom_node *node);
 /* Returns the previous sibling to the node. */
 struct dom_node *get_dom_node_prev(struct dom_node *node);
 
+/* Returns the next sibling to the node. */
+struct dom_node *get_dom_node_next(struct dom_node *node);
+
 /* Returns first text node of the element or NULL. */
 struct dom_node *
 get_dom_node_child(struct dom_node *node, enum dom_node_type child_type,

src/dom/sgml/Makefile

@@ -2,6 +2,6 @@ top_builddir=../../..
 include $(top_builddir)/Makefile.config
 
 SUBDIRS = docbook html rss xbel
-OBJS	= sgml.o parser.o scanner.o
+OBJS	= dump.o parser.o scanner.o sgml.o
 
 include $(top_srcdir)/Makefile.lib

src/dom/sgml/dump.c

@@ -0,0 +1,170 @@
+/* SGML generics */
+
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <stdlib.h>
+#include <string.h>
+
+#include "elinks.h"
+
+#include "dom/node.h"
+#include "dom/sgml/dump.h"
+#include "dom/stack.h"
+#include "dom/string.h"
+
+
+static enum dom_stack_code
+sgml_file_dumper_element_push(struct dom_stack *stack, struct dom_node *node, void *data)
+{
+	FILE *file = stack->current->data;
+	struct dom_string *string = &node->string;
+
+	fprintf(file, "<%.*s", string->length, string->string);
+	if (!node->data.element.map
+	    || node->data.element.map->size == 0)
+		fprintf(file, ">");
+
+	return DOM_STACK_CODE_OK;
+}
+
+static enum dom_stack_code
+sgml_file_dumper_element_pop(struct dom_stack *stack, struct dom_node *node, void *data)
+{
+	FILE *file = stack->current->data;
+	struct dom_string *string = &node->string;
+
+	fprintf(file, "</%.*s>", string->length, string->string);
+
+	return DOM_STACK_CODE_OK;
+}
+
+static enum dom_stack_code
+sgml_file_dumper_attribute_push(struct dom_stack *stack, struct dom_node *node, void *data)
+{
+	FILE *file = stack->current->data;
+	struct dom_string *name = &node->string;
+	struct dom_string *value = &node->data.attribute.value;
+
+	if (node->parent->type == DOM_NODE_PROCESSING_INSTRUCTION)
+		return DOM_STACK_CODE_OK;
+
+	fprintf(file, " %.*s", name->length, name->string);
+
+	if (value->string) {
+		if (node->data.attribute.quoted)
+			fprintf(file, "=%c%.*s%c",
+				node->data.attribute.quoted,
+				value->length, value->string,
+				node->data.attribute.quoted);
+		else
+			/* FIXME: 'encode' if the value contains '"'s? */
+			fprintf(file, "=\"%.*s\"", value->length, value->string);
+	}
+
+	if (!get_dom_node_next(node))
+		fprintf(file, ">");
+
+	return DOM_STACK_CODE_OK;
+}
+
+static enum dom_stack_code
+sgml_file_dumper_proc_instruction_push(struct dom_stack *stack, struct dom_node *node, void *data)
+{
+	FILE *file = stack->current->data;
+	struct dom_string *target = &node->string;
+	struct dom_string *instruction = &node->data.proc_instruction.instruction;
+
+	fprintf(file, "<?%.*s %.*s?>",
+		target->length, target->string,
+		instruction->length, instruction->string);
+
+	return DOM_STACK_CODE_OK;
+}
+
+static enum dom_stack_code
+sgml_file_dumper_text_push(struct dom_stack *stack, struct dom_node *node, void *data)
+{
+	FILE *file = stack->current->data;
+	struct dom_string *string = &node->string;
+
+	fprintf(file, "%.*s", string->length, string->string);
+
+	return DOM_STACK_CODE_OK;
+}
+
+static enum dom_stack_code
+sgml_file_dumper_cdata_section_push(struct dom_stack *stack, struct dom_node *node, void *data)
+{
+	FILE *file = stack->current->data;
+	struct dom_string *string = &node->string;
+
+	fprintf(file, "<![CDATA[%.*s]]>", string->length, string->string);
+
+	return DOM_STACK_CODE_OK;
+}
+
+static enum dom_stack_code
+sgml_file_dumper_comment_push(struct dom_stack *stack, struct dom_node *node, void *data)
+{
+	FILE *file = stack->current->data;
+	struct dom_string *string = &node->string;
+
+	fprintf(file, "<!--%.*s-->", string->length, string->string);
+
+	return DOM_STACK_CODE_OK;
+}
+
+static enum dom_stack_code
+sgml_file_dumper_entity_ref_push(struct dom_stack *stack, struct dom_node *node, void *data)
+{
+	FILE *file = stack->current->data;
+	struct dom_string *string = &node->string;
+
+	fprintf(file, "&%.*s;", string->length, string->string);
+
+	return DOM_STACK_CODE_OK;
+}
+
+struct dom_stack_context_info sgml_file_dumper = {
+	/* Object size: */			0,
+	/* Push: */
+	{
+		/*				*/ NULL,
+		/* DOM_NODE_ELEMENT		*/ sgml_file_dumper_element_push,
+		/* DOM_NODE_ATTRIBUTE		*/ sgml_file_dumper_attribute_push,
+		/* DOM_NODE_TEXT		*/ sgml_file_dumper_text_push,
+		/* DOM_NODE_CDATA_SECTION	*/ sgml_file_dumper_cdata_section_push,
+		/* DOM_NODE_ENTITY_REFERENCE	*/ sgml_file_dumper_entity_ref_push,
+		/* DOM_NODE_ENTITY		*/ NULL,
+		/* DOM_NODE_PROC_INSTRUCTION	*/ sgml_file_dumper_proc_instruction_push,
+		/* DOM_NODE_COMMENT		*/ sgml_file_dumper_comment_push,
+		/* DOM_NODE_DOCUMENT		*/ NULL,
+		/* DOM_NODE_DOCUMENT_TYPE	*/ NULL,
+		/* DOM_NODE_DOCUMENT_FRAGMENT	*/ NULL,
+		/* DOM_NODE_NOTATION		*/ NULL,
+	},
+	/* Pop: */
+	{
+		/*				*/ NULL,
+		/* DOM_NODE_ELEMENT		*/ sgml_file_dumper_element_pop,
+		/* DOM_NODE_ATTRIBUTE		*/ NULL,
+		/* DOM_NODE_TEXT		*/ NULL,
+		/* DOM_NODE_CDATA_SECTION	*/ NULL,
+		/* DOM_NODE_ENTITY_REFERENCE	*/ NULL,
+		/* DOM_NODE_ENTITY		*/ NULL,
+		/* DOM_NODE_PROC_INSTRUCTION	*/ NULL,
+		/* DOM_NODE_COMMENT		*/ NULL,
+		/* DOM_NODE_DOCUMENT		*/ NULL,
+		/* DOM_NODE_DOCUMENT_TYPE	*/ NULL,
+		/* DOM_NODE_DOCUMENT_FRAGMENT	*/ NULL,
+		/* DOM_NODE_NOTATION		*/ NULL,
+	}
+};
+
+struct dom_stack_context *
+add_sgml_file_dumper(struct dom_stack *stack, FILE *file)
+{
+	return add_dom_stack_context(stack, file, &sgml_file_dumper);
+}

src/dom/sgml/dump.h

@@ -0,0 +1,12 @@
+#ifndef EL_DOM_SGML_DUMP_H
+#define EL_DOM_SGML_DUMP_H
+
+#include <stdio.h>
+
+struct dom_stack;
+struct dom_stack_context;
+
+struct dom_stack_context *
+add_sgml_file_dumper(struct dom_stack *stack, FILE *file);
+
+#endif

src/dom/sgml/parser.c

@@ -95,7 +95,7 @@ add_sgml_attribute(struct dom_stack *stack,
 	node->data.attribute.reference = !!(info->flags & SGML_ATTRIBUTE_REFERENCE);
 
 	if (valtoken && valtoken->type == SGML_TOKEN_STRING)
-		node->data.attribute.quoted = 1;
+		node->data.attribute.quoted = valtoken->string.string[-1];
 
 	if (!node || push_dom_node(stack, node) != DOM_STACK_CODE_OK)
 		return NULL;

src/dom/test/sgml-parser.c

@@ -13,6 +13,7 @@
 
 #include "dom/configuration.h"
 #include "dom/node.h"
+#include "dom/sgml/dump.h"
 #include "dom/sgml/parser.h"
 #include "dom/stack.h"
 
@@ -265,6 +266,7 @@ main(int argc, char *argv[])
 	enum sgml_parser_code code = 0;
 	enum dom_config_flag normalize_flags = 0;
 	int normalize = 0;
+	int dump = 0;
 	int complete = 1;
 	size_t read_stdin = 0;
 	struct dom_string uri = STATIC_DOM_STRING("dom://test");
@@ -338,6 +340,10 @@ main(int argc, char *argv[])
 			flags |= SGML_PARSER_INCREMENTAL;
 			complete = 0;
 
+		} else if (!strcmp(arg, "dump")) {
+			type = SGML_PARSER_TREE;
+			dump = 1;
+
 		} else if (!strcmp(arg, "error")) {
 			flags |= SGML_PARSER_DETECT_ERRORS;
 
@@ -355,7 +361,7 @@ main(int argc, char *argv[])
 	parser->error_func = sgml_error_function;
 	if (normalize)
 		add_dom_config_normalizer(&parser->stack, normalize_flags);
-	else
+	else if (!dump)
 		add_dom_stack_context(&parser->stack, NULL, &sgml_parser_test_context_info);
 
 	if (read_stdin > 0) {
@@ -402,7 +408,7 @@ main(int argc, char *argv[])
 			pop_dom_node(&parser->stack);
 		}
 
-		if (normalize) {
+		if (normalize || dump) {
 			struct dom_stack stack;
 
 			/* Note, that we cannot free nodes when walking the DOM
@@ -412,7 +418,10 @@ main(int argc, char *argv[])
 			/* XXX: This context needs to be added first because it
 			 * assumes the parser can be accessed via
 			 * stack->contexts[0].data. */
-			add_dom_stack_context(&stack, parser, &sgml_parser_test_context_info);
+			if (normalize)
+				add_dom_stack_context(&stack, parser, &sgml_parser_test_context_info);
+			else if (dump)
+				add_sgml_file_dumper(&stack, stdout);
 			walk_dom_nodes(&stack, parser->root);
 			done_dom_stack(&stack);
 			done_dom_node(parser->root);

src/dom/test/test-sgml-dump-basic

@@ -0,0 +1,173 @@
+#!/bin/sh
+#
+# Copyright (c) 2005 Jonas Fonseca
+#
+
+test_description='Test dumping of SGML documents.
+
+Test that DOM documents are consistently dumped to SGML.
+Note, this also test whether attribute nodes are sorted
+correctly.
+'
+
+. "$TEST_LIB"
+
+test_sgml_dump_exact () {
+	desc="$1"; shift
+	src="$1"; shift
+
+	sgml-parser --dump --src "$src" > output
+	echo -n "$src" > expected
+
+	test_expect_success "$desc" 'cmp output expected' 
+}
+
+test_sgml_dump_equals () {
+	desc="$1"; shift
+	src="$1"; shift
+	out="$1"; shift
+
+	sgml-parser --dump --src "$src" > output
+	echo -n "$out" > expected
+
+	test_expect_success "$desc" 'cmp output expected' 
+}
+
+
+################################################################
+# Parse various SGML node types.
+
+test_sgml_dump_exact \
+'Parse a small document.' \
+'<html><body><p>Hello World!</p></body></html>'
+
+test_sgml_dump_exact \
+'Parse an enclosed comment.' \
+'<root><!-- Hello World! --></root>'
+
+test_sgml_dump_exact \
+'Parse comment combinations. (I)' \
+'<root><!-- <!-- -- > --><!--foo--><!----></root>'
+
+test_sgml_dump_exact \
+'Parse comment combinations. (II).' \
+'<!-- comment -->s<!-->-->t<!----->u'
+
+test_sgml_dump_exact \
+'Parse empty comment.' \
+'<!---->s'
+
+test_sgml_dump_exact \
+'Parse an enclosed CDATA section.' \
+'<root><![CDATA[...] ]>...]]></root>'
+
+test_sgml_dump_exact \
+'Parse non-enclosed CDATA section.' \
+'<![CDATA[...]]>'
+
+test_sgml_dump_exact \
+'Parse attributes.' \
+'<e a="1" b="2" c=""></e>'
+
+test_sgml_dump_exact \
+'Parse XML stylesheet processing instructions.' \
+'<?xml-stylesheet type="text/xsl" href="url"?>'
+
+test_sgml_dump_exact \
+'Parse entity references.' \
+'&amp;-&#42;'
+
+#############################################################################
+# Test tidy up dumping
+
+test_sgml_dump_equals \
+'Parse elements.' \
+'<root><child attr="value" /><child2></><child3 >a</></root>' \
+'<root><child attr="value"></child><child2></child2><child3>a</child3></root>'
+
+test_sgml_dump_equals \
+'Parse attributes with garbage.' \
+"<root a=b c='d' e'f' g= h i = j k =></root>" \
+"<root a=\"b\" c='d' g=\"h\" i=\"j\" k></root>"
+
+test_sgml_dump_equals \
+'Parse bad comment. (II)' \
+'<!--a--!>bad comment' \
+'<!--a-->bad comment'
+
+test_sgml_dump_equals \
+'Parse empty notation.' \
+'<!>s' \
+'s'
+
+test_sgml_dump_equals \
+'Parse a bad CDATA section.' \
+'<![CDATA[...' \
+'<![CDATA[...]]>'
+
+test_sgml_dump_equals \
+'Parse tag soup elements. (I)' \
+'<parent attr="value" <child:1></><child:2</>a</parent>' \
+'<parent attr="value"><child:1></child:1><child:2></child:2>a</parent>'
+
+test_sgml_dump_equals \
+'Parse tag soup elements. (II)' \
+'< a >< b < c / >< / >' \
+'<a><b><c></c></b></a>'
+
+test_sgml_dump_equals \
+'Parse attribute with non-quoted values.' \
+'<root color=#abc path=/to/%61-&\one";files/>...' \
+'<root color="#abc" path="/to/%61-&\one";files"></root>...'
+
+# Just how these should be gracefully handled is not clear to me.
+test_sgml_dump_equals \
+'Parse badly formatted entity references.' \
+'& m33p;-&.:-copy;-&;-&#;-&#xx;' \
+'& m33p;-&.:-copy;-&;-&#;-&#xx;'
+
+test_sgml_dump_equals \
+'Parse processing instructions.' \
+'<?xml encoding="UTF8"?>
+...
+<?ecmascript
+var val=2;
+?>' \
+'<?xml encoding="UTF8"?>
+...
+<?ecmascript var val=2;
+?>'
+
+test_sgml_dump_equals \
+'Parse XML processing instructions.' \
+'<?xml version="1.0" />?><?xml />-' \
+'<?xml version="1.0" />?><?xml />-?>'
+
+test_sgml_dump_equals \
+'Parse exotic processing instructions.' \
+'<?xml ?+>+?>-?>-<?js?>-<??>-' \
+'<?xml ?+>+?>-?>-<?js ?>-<? ?>-'
+
+test_sgml_dump_equals \
+'Parse incorrect processing instructions. (I)' \
+'<?js<?>-<?<??>-<?xml <=";&?>-<?' \
+'<?js <?>-<? <??>-<?xml <=";&?>-' \
+
+test_sgml_dump_equals \
+'Parse incorrect processing instructions. (II)' \
+'<?><?' \
+'<? ><??>'
+
+test_sgml_dump_equals \
+'Skip spaces not inside text.' \
+'<
+root
+ns:attr                      
+=
+"value"
+><?	
+	target	
+ data?><	/	root	>' \
+'<root ns:attr="value"><?target data?></root>'
+
+test_done

src/dom/test/test-sgml-parser-basic

@@ -225,7 +225,7 @@ proc-instruction:  ->
 #text: -'
 
 test_output_equals \
-'Parse incorrect processing instructions.' \
+'Parse incorrect processing instructions. (I)' \
 '<?js<?>-<?<??>-<?xml <=";&?>-<?' \
 '
 proc-instruction: js -> <

src/protocol/Makefile

@@ -1,8 +1,6 @@
 top_builddir=../..
 include $(top_builddir)/Makefile.config
 
-SUBDIRS = auth file http
-
 SUBDIRS-$(CONFIG_BITTORRENT)	+= bittorrent
 SUBDIRS-$(CONFIG_FINGER)	+= finger
 SUBDIRS-$(CONFIG_FSP)		+= fsp
@@ -12,11 +10,10 @@ SUBDIRS-$(CONFIG_NNTP)		+= nntp
 SUBDIRS-$(CONFIG_SMB)		+= smb
 SUBDIRS-$(CONFIG_URI_REWRITE)	+= rewrite
 
-OBJS = about.o date.o header.o protocol.o proxy.o uri.o user.o
+SUBDIRS = auth file http
 
 OBJS-$(CONFIG_DATA)		+= data.o
-OBJS-$(CONFIG_CGI)		+= common.o
-OBJS-$(CONFIG_FSP)		+= common.o
-OBJS-$(CONFIG_SMB)		+= common.o
+
+OBJS = about.o common.o date.o header.o protocol.o proxy.o uri.o user.o
 
 include $(top_srcdir)/Makefile.lib

src/protocol/common.c

@@ -17,7 +17,13 @@
 
 #include "elinks.h"
 
+#include "config/options.h"
 #include "protocol/common.h"
+#include "protocol/protocol.h"
+#include "protocol/uri.h"
+#include "util/conv.h"
+#include "util/memory.h"
+#include "util/string.h"
 
 
 /* Close all non-terminal file descriptors. */
@@ -35,3 +41,103 @@ close_all_non_term_fd(void)
 	for (n = 3; n < max; n++)
 		close(n);
 }
+
+enum connection_state
+init_directory_listing(struct string *page, struct uri *uri)
+{
+	struct string dirpath = NULL_STRING;
+	struct string location = NULL_STRING;
+	unsigned char *info;
+	int local = (uri->protocol == PROTOCOL_FILE);
+
+	if (!init_string(page)
+	    || !init_string(&dirpath)
+	    || !init_string(&location)
+	    || !add_uri_to_string(&dirpath, uri, URI_DATA)
+	    || !add_uri_to_string(&location, uri, URI_DIR_LOCATION))
+		goto out_of_memory;
+
+	if (dirpath.length > 0
+	    && !dir_sep(dirpath.source[dirpath.length - 1])
+	    && !add_char_to_string(&dirpath, '/'))
+		goto out_of_memory;
+
+	if (uri->protocol == PROTOCOL_GOPHER) {
+		/* A little hack to get readable Gopher names. We should find a
+		 * way to do it more general. */
+		decode_uri_string(&dirpath);
+	}
+
+	if (!local && !add_char_to_string(&location, '/'))
+		goto out_of_memory;
+
+	if (!add_to_string(page, "<html>\n<head><title>"))
+		goto out_of_memory;
+
+	if (!local && !add_string_to_string(page, &location))
+		goto out_of_memory;
+
+	if (!add_html_to_string(page, dirpath.source, dirpath.length)
+	    || !add_to_string(page, "</title>\n<base href=\"")
+	    || !add_string_to_string(page, &location))
+		goto out_of_memory;
+
+	encode_uri_string(page, dirpath.source, -1, 0);
+
+	if (!add_to_string(page, "\" />\n</head>\n<body>\n<h2>"))
+		goto out_of_memory;
+
+	/* Use module names? */
+	switch (uri->protocol) {
+	case PROTOCOL_FILE:
+		info = "Local";
+		break;
+	case PROTOCOL_FSP:
+		info = "FSP";
+		break;
+	case PROTOCOL_FTP:
+		info = "FTP";
+		break;
+	case PROTOCOL_GOPHER:
+		info = "Gopher";
+		break;
+	default:
+		info = "?";
+	}
+
+	if (!add_to_string(page, info)
+	    || !add_to_string(page, " directory "))
+		goto out_of_memory;
+
+	if (!local && !add_string_to_string(page, &location))
+		goto out_of_memory;
+
+	/* Make the directory path with links to each subdir. */
+	{
+		unsigned char *slash = dirpath.source;
+		unsigned char *pslash = slash;
+
+		while ((slash = strchr(slash, '/'))) {
+			/* FIXME: htmlesc? At least we should escape quotes. --pasky */
+			if (!add_to_string(page, "<a href=\"")
+	    		    || !add_string_to_string(page, &location)
+			    || !add_bytes_to_string(page, dirpath.source, slash - dirpath.source)
+			    || !add_to_string(page, "/\">")
+			    || !add_html_to_string(page, pslash, slash - pslash)
+			    || !add_to_string(page, "</a>/"))
+				goto out_of_memory;
+
+			pslash = ++slash;
+		}
+	}
+
+	if (!add_to_string(page, "</h2>\n<pre>")) {
+out_of_memory:
+		done_string(page);
+	}
+
+	done_string(&dirpath);
+	done_string(&location);
+
+	return page->length > 0 ? S_OK : S_OUT_OF_MEM;
+}

src/protocol/common.h

@@ -1,7 +1,15 @@
 #ifndef EL__PROTOCOL_COMMON_H
 #define EL__PROTOCOL_COMMON_H
 
+#include "network/state.h"
+
+struct string;
+struct uri;
+
 /* Close all non-terminal file descriptors. */
 void close_all_non_term_fd(void);
 
+enum connection_state
+init_directory_listing(struct string *page, struct uri *uri);
+
 #endif

src/protocol/file/file.c

@@ -25,6 +25,7 @@
 #include "intl/gettext/libintl.h"
 #include "main/module.h"
 #include "network/connection.h"
+#include "protocol/common.h"
 #include "protocol/file/cgi.h"
 #include "protocol/file/file.h"
 #include "protocol/uri.h"
@@ -172,12 +173,12 @@ add_dir_entries(struct directory_entry *entries, unsigned char *dirpath,
  * @dirpath. */
 /* Returns a connection state. S_OK if all is well. */
 static inline enum connection_state
-list_directory(unsigned char *dirpath, struct string *page)
+list_directory(struct connection *conn, unsigned char *dirpath,
+	       struct string *page)
 {
 	int show_hidden_files = get_opt_bool("protocol.file.show_hidden_files");
-	unsigned char *slash = dirpath;
-	unsigned char *pslash = ++slash;
 	struct directory_entry *entries;
+	enum connection_state state;
 
 	errno = 0;
 	entries = get_directory_entries(dirpath, show_hidden_files);
@@ -186,30 +187,17 @@ list_directory(unsigned char *dirpath, struct string *page)
 		return S_OUT_OF_MEM;
 	}
 
-	if (!init_string(page)) return S_OUT_OF_MEM;
+	state = init_directory_listing(page, conn->uri);
+	if (state != S_OK)
+		return S_OUT_OF_MEM;
 
-	add_to_string(page, "<html>\n<head><title>");
-	add_html_to_string(page, dirpath, strlen(dirpath));
-	add_to_string(page, "</title>\n<base href=\"");
-	encode_uri_string(page, dirpath, -1, 0);
-	add_to_string(page, "\" />\n</head>\n<body>\n<h2>Directory /");
+	add_dir_entries(entries, dirpath, page);
 
-	/* Make the directory path with links to each subdir. */
-	while ((slash = strchr(slash, '/'))) {
-		*slash = 0;
-		add_to_string(page, "<a href=\"");
-		/* FIXME: htmlesc? At least we should escape quotes. --pasky */
-		add_to_string(page, dirpath);
-		add_to_string(page, "/\">");
-		add_html_to_string(page, pslash, strlen(pslash));
-		add_to_string(page, "</a>/");
-		*slash = '/';
-		pslash = ++slash;
+	if (!add_to_string(page, "</pre>\n<hr>\n</body>\n</html>\n")) {
+		done_string(page);
+		return S_OUT_OF_MEM;
 	}
 
-	add_to_string(page, "</h2>\n<pre>");
-	add_dir_entries(entries, dirpath, page);
-	add_to_string(page, "</pre>\n<hr>\n</body>\n</html>\n");
 	return S_OK;
 }
 
@@ -263,7 +251,7 @@ file_protocol_handler(struct connection *connection)
 			redirect_location = "/";
 			state = S_OK;
 		} else {
-			state = list_directory(name.source, &page);
+			state = list_directory(connection, name.source, &page);
 			type = "text/html";
 		}
 

src/protocol/fsp/fsp.c

@@ -131,21 +131,16 @@ fsp_directory(FSP_SESSION *ses, struct uri *uri)
 {
 	struct string buf;
 	FSP_DIR *dir;
-	unsigned char *uristring = get_uri_string(uri, URI_PUBLIC);
 	unsigned char *data = get_uri_string(uri, URI_DATA);
 	unsigned char dircolor[8] = "";
 
-	if (!uristring || !data || !init_string(&buf))
+	if (!data || init_directory_listing(&buf, uri) != S_OK)
 		fsp_error("Out of memory");
 
 	fprintf(stderr, "text/html");
 	fclose(stderr);
-	add_html_to_string(&buf, uristring, strlen(uristring));
 
-	printf("<html><head><title>%s</title><base href=\"%s", buf.source,
-		uristring);
-	if (buf.source[buf.length - 1] != '/') printf("/");
-	printf("\"></head><body><h2>FSP directory %s</h2><pre>", buf.source);
+	puts(buf.source);
 
 	dir = fsp_opendir(ses, data);
 	if (!dir) goto end;

src/protocol/ftp/ftp.c

@@ -42,6 +42,7 @@
 #include "osdep/osdep.h"
 #include "osdep/stat.h"
 #include "protocol/auth/auth.h"
+#include "protocol/common.h"
 #include "protocol/ftp/ftp.h"
 #include "protocol/ftp/parse.h"
 #include "protocol/uri.h"
@@ -1253,37 +1254,21 @@ out_of_mem:
 
 	if (ftp->dir && !conn->from) {
 		struct string string;
-		unsigned char *uristring;
+		enum connection_state state;
 
 		if (!conn->uri->data) {
 			abort_connection(conn, S_FTP_ERROR);
 			return;
 		}
 
-		uristring = get_uri_string(conn->uri, URI_PUBLIC);
-		if (!uristring) goto out_of_mem;
-
-		if (!init_string(&string)) {
-			mem_free(uristring);
-			goto out_of_mem;
+		state = init_directory_listing(&string, conn->uri);
+		if (state != S_OK) {
+			abort_connection(conn, state);
+			return;
 		}
 
-		add_html_to_string(&string, uristring, strlen(uristring));
-		mem_free(uristring);
-
-#define ADD_CONST(str) { \
-	add_fragment(conn->cached, conn->from, str, sizeof(str) - 1); \
-	conn->from += (sizeof(str) - 1); }
-
-#define ADD_STRING() { \
-	add_fragment(conn->cached, conn->from, string.source, string.length); \
-	conn->from += string.length; }
-
-		ADD_CONST("<html>\n<head><title>");
-		ADD_STRING();
-		ADD_CONST("</title></head>\n<body>\n<h2>FTP directory ");
-		ADD_STRING();
-		ADD_CONST("</h2>\n<pre>");
+		add_fragment(conn->cached, conn->from, string.source, string.length);
+		conn->from += string.length;
 
 		done_string(&string);
 
@@ -1343,6 +1328,10 @@ out_of_mem:
 				(unsigned char *) dircolor) == -1)
 		goto out_of_mem;
 
+#define ADD_CONST(str) { \
+	add_fragment(conn->cached, conn->from, str, sizeof(str) - 1); \
+	conn->from += (sizeof(str) - 1); }
+
 	if (ftp->dir) ADD_CONST("</pre>\n<hr>\n</body>\n</html>");
 
 	close_socket(conn->data_socket);

src/protocol/gopher/gopher.c

@@ -35,6 +35,7 @@
 #include "main/module.h"
 #include "network/connection.h"
 #include "network/socket.h"
+#include "protocol/common.h"
 #include "protocol/gopher/gopher.h"
 #include "protocol/protocol.h"
 #include "protocol/uri.h"
@@ -599,25 +600,15 @@ read_gopher_directory_data(struct connection *conn, struct read_buffer *rb)
 	struct string buffer;
 	unsigned char *end;
 
-	if (!init_string(&buffer))
-		return S_OUT_OF_MEM;
-
 	if (conn->from == 0) {
-		unsigned char *where = get_uri_string(conn->uri, URI_PUBLIC);
+		enum connection_state state;
 
-		if (where) decode_uri_for_display(where);
+		state = init_directory_listing(&buffer, conn->uri);
+		if (state != S_OK)
+			return state;
 
-		add_format_to_string(&buffer,
-			"<html>\n"
-			"<head>\n"
-			"<title>Gopher menu at %s</title>\n"
-			"</head>\n"
-			"<body>\n"
-			"<h1>Gopher menu at %s</h1>\n"
-			"<pre>",
-			empty_string_or_(where), empty_string_or_(where));
-
-		mem_free_if(where);
+	} else if (!init_string(&buffer)) {
+		return S_OUT_OF_MEM;
 	}
 
 	while ((end = get_gopher_line_end(rb->data, rb->length))) {

src/protocol/uri.h

@@ -160,6 +160,9 @@ enum uri_component {
 	/* Used for HTTP CONNECT method info */
 	URI_HTTP_CONNECT	= URI_HOST | URI_PORT | URI_DEFAULT_PORT,
 
+	/* Used for adding directory listing HTML header, */
+	URI_DIR_LOCATION	= URI_PROTOCOL | URI_HOST | URI_PORT | URI_IDN,
+
 	/* Used for getting the host of a DNS query. As a hidden bonus we get
 	 * IPv6 hostnames without the brackets because we don't ask for
 	 * URI_PORT. */