/* HTML parser for Wget.
- Copyright (C) 1998, 2000, 2003 Free Software Foundation, Inc.
+ Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
+ 2007, 2008, 2009, 2010, 2011 Free Software Foundation, Inc.
This file is part of GNU Wget.
GNU Wget is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
-the Free Software Foundation; either version 2 of the License, or (at
+the Free Software Foundation; either version 3 of the License, or (at
your option) any later version.
GNU Wget is distributed in the hope that it will be useful,
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
-along with Wget; if not, write to the Free Software
-Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
-
-In addition, as a special exception, the Free Software Foundation
-gives permission to link the code of its release of Wget with the
-OpenSSL project's "OpenSSL" library (or with modified versions of it
-that use the same license as the "OpenSSL" library), and distribute
-the linked executables. You must obey the GNU General Public License
-in all respects for all of the code used other than "OpenSSL". If you
-modify this file, you may extend this exception to your version of the
-file, but you are not obligated to do so. If you do not wish to do
-so, delete this exception statement from your version. */
+along with Wget. If not, see <http://www.gnu.org/licenses/>.
+
+Additional permission under GNU GPL version 3 section 7
+
+If you modify this program, or any covered work, by linking or
+combining it with the OpenSSL project's OpenSSL library (or a
+modified version of that library), containing parts covered by the
+terms of the OpenSSL or SSLeay licenses, the Free Software Foundation
+grants you additional permission to convey the resulting work.
+Corresponding Source for a non-source form of such a combination
+shall include the source code for the parts of OpenSSL used as well
+as that of the covered work. */
/* The only entry point to this module is map_html_tags(), which see. */
/* To test as standalone, compile with `-DSTANDALONE -I.'. You'll
still need Wget headers to compile. */
-#include <config.h>
+#include "wget.h"
#ifdef STANDALONE
# define I_REALLY_WANT_CTYPE_MACROS
#include <stdio.h>
#include <stdlib.h>
-#ifdef HAVE_STRING_H
-# include <string.h>
-#else
-# include <strings.h>
-#endif
+#include <string.h>
#include <assert.h>
-#include "wget.h"
+#include "utils.h"
#include "html-parse.h"
#ifdef STANDALONE
# define xrealloc realloc
# define xfree free
-# undef ISSPACE
-# undef ISDIGIT
-# undef ISXDIGIT
-# undef ISALPHA
-# undef ISALNUM
-# undef TOLOWER
-# undef TOUPPER
-
-# define ISSPACE(x) isspace (x)
-# define ISDIGIT(x) isdigit (x)
-# define ISXDIGIT(x) isxdigit (x)
-# define ISALPHA(x) isalpha (x)
-# define ISALNUM(x) isalnum (x)
-# define TOLOWER(x) tolower (x)
-# define TOUPPER(x) toupper (x)
-
-static struct options opt;
-#endif /* STANDALONE */
+# undef c_isspace
+# undef c_isdigit
+# undef c_isxdigit
+# undef c_isalpha
+# undef c_isalnum
+# undef c_tolower
+# undef c_toupper
+
+# define c_isspace(x) isspace (x)
+# define c_isdigit(x) isdigit (x)
+# define c_isxdigit(x) isxdigit (x)
+# define c_isalpha(x) isalpha (x)
+# define c_isalnum(x) isalnum (x)
+# define c_tolower(x) tolower (x)
+# define c_toupper(x) toupper (x)
+
+struct hash_table {
+ int dummy;
+};
+static void *
+hash_table_get (const struct hash_table *ht, void *ptr)
+{
+ return ptr;
+}
+#else /* not STANDALONE */
+# include "hash.h"
+#endif
/* Pool support. A pool is a resizable chunk of memory. It is first
allocated on the stack, and moved to the heap if it needs to be
allocation because the entire pool is kept on the stack. */
struct pool {
- char *contents; /* pointer to the contents. */
- int size; /* size of the pool. */
- int tail; /* next available position index. */
- int resized; /* whether the pool has been resized
- using malloc. */
+ char *contents; /* pointer to the contents. */
+ int size; /* size of the pool. */
+ int tail; /* next available position index. */
+ bool resized; /* whether the pool has been resized
+ using malloc. */
- char *orig_contents; /* original pool contents, usually
+ char *orig_contents; /* original pool contents, usually
stack-allocated. used by POOL_FREE
to restore the pool to the initial
state. */
/* Initialize the pool to hold INITIAL_SIZE bytes of storage. */
-#define POOL_INIT(p, initial_storage, initial_size) do { \
- struct pool *P = (p); \
- P->contents = (initial_storage); \
- P->size = (initial_size); \
- P->tail = 0; \
- P->resized = 0; \
- P->orig_contents = P->contents; \
- P->orig_size = P->size; \
+#define POOL_INIT(p, initial_storage, initial_size) do { \
+ struct pool *P = (p); \
+ P->contents = (initial_storage); \
+ P->size = (initial_size); \
+ P->tail = 0; \
+ P->resized = false; \
+ P->orig_contents = P->contents; \
+ P->orig_size = P->size; \
} while (0)
/* Grow the pool to accomodate at least SIZE new bytes. If the pool
already has room to accomodate SIZE bytes of data, this is a no-op. */
-#define POOL_GROW(p, increase) \
- GROW_ARRAY ((p)->contents, (p)->size, (p)->tail + increase, \
- (p)->resized, char)
+#define POOL_GROW(p, increase) \
+ GROW_ARRAY ((p)->contents, (p)->size, (p)->tail + (increase), \
+ (p)->resized, char)
/* Append text in the range [beg, end) to POOL. No zero-termination
is done. */
-#define POOL_APPEND(p, beg, end) do { \
- const char *PA_beg = (beg); \
- int PA_size = (end) - PA_beg; \
- POOL_GROW (p, PA_size); \
- memcpy ((p)->contents + (p)->tail, PA_beg, PA_size); \
- (p)->tail += PA_size; \
+#define POOL_APPEND(p, beg, end) do { \
+ const char *PA_beg = (beg); \
+ int PA_size = (end) - PA_beg; \
+ POOL_GROW (p, PA_size); \
+ memcpy ((p)->contents + (p)->tail, PA_beg, PA_size); \
+ (p)->tail += PA_size; \
} while (0)
/* Append one character to the pool. Can be used to zero-terminate
pool strings. */
-#define POOL_APPEND_CHR(p, ch) do { \
- char PAC_char = (ch); \
- POOL_GROW (p, 1); \
- (p)->contents[(p)->tail++] = PAC_char; \
+#define POOL_APPEND_CHR(p, ch) do { \
+ char PAC_char = (ch); \
+ POOL_GROW (p, 1); \
+ (p)->contents[(p)->tail++] = PAC_char; \
} while (0)
/* Forget old pool contents. The allocated memory is not freed. */
values. That way after POOL_FREE, the pool is fully usable, just
as if it were freshly initialized with POOL_INIT. */
-#define POOL_FREE(p) do { \
- struct pool *P = p; \
- if (P->resized) \
- xfree (P->contents); \
- P->contents = P->orig_contents; \
- P->size = P->orig_size; \
- P->tail = 0; \
- P->resized = 0; \
+#define POOL_FREE(p) do { \
+ struct pool *P = p; \
+ if (P->resized) \
+ xfree (P->contents); \
+ P->contents = P->orig_contents; \
+ P->size = P->orig_size; \
+ P->tail = 0; \
+ P->resized = false; \
} while (0)
/* Used for small stack-allocated memory chunks that might grow. Like
After the first resize, subsequent ones are performed with realloc,
just like DO_REALLOC. */
-#define GROW_ARRAY(basevar, sizevar, needed_size, resized, type) do { \
- long ga_needed_size = (needed_size); \
- long ga_newsize = (sizevar); \
- while (ga_newsize < ga_needed_size) \
- ga_newsize <<= 1; \
- if (ga_newsize != (sizevar)) \
- { \
- if (resized) \
- basevar = (type *)xrealloc (basevar, ga_newsize * sizeof (type)); \
- else \
- { \
- void *ga_new = xmalloc (ga_newsize * sizeof (type)); \
- memcpy (ga_new, basevar, (sizevar) * sizeof (type)); \
- (basevar) = ga_new; \
- resized = 1; \
- } \
- (sizevar) = ga_newsize; \
- } \
+#define GROW_ARRAY(basevar, sizevar, needed_size, resized, type) do { \
+ long ga_needed_size = (needed_size); \
+ long ga_newsize = (sizevar); \
+ while (ga_newsize < ga_needed_size) \
+ ga_newsize <<= 1; \
+ if (ga_newsize != (sizevar)) \
+ { \
+ if (resized) \
+ basevar = xrealloc (basevar, ga_newsize * sizeof (type)); \
+ else \
+ { \
+ void *ga_new = xmalloc (ga_newsize * sizeof (type)); \
+ memcpy (ga_new, basevar, (sizevar) * sizeof (type)); \
+ (basevar) = ga_new; \
+ resized = true; \
+ } \
+ (sizevar) = ga_newsize; \
+ } \
} while (0)
\f
-#define AP_DOWNCASE 1
-#define AP_PROCESS_ENTITIES 2
-#define AP_TRIM_BLANKS 4
+/* Test whether n+1-sized entity name fits in P. We don't support
+ IE-style non-terminated entities, e.g. "<foo" -> "<foo".
+ However, "<foo" will work, as will "<!foo", "<", etc. In
+ other words an entity needs to be terminated by either a
+ non-alphanumeric or the end of string. */
+#define FITS(p, n) (p + n == end || (p + n < end && !c_isalnum (p[n])))
+
+/* Macros that test entity names by returning true if P is followed by
+ the specified characters. */
+#define ENT1(p, c0) (FITS (p, 1) && p[0] == c0)
+#define ENT2(p, c0, c1) (FITS (p, 2) && p[0] == c0 && p[1] == c1)
+#define ENT3(p, c0, c1, c2) (FITS (p, 3) && p[0]==c0 && p[1]==c1 && p[2]==c2)
+
+/* Increment P by INC chars. If P lands at a semicolon, increment it
+ past the semicolon. This ensures that e.g. "<foo" is converted
+ to "<foo", but "<,foo" to "<,foo". */
+#define SKIP_SEMI(p, inc) (p += inc, p < end && *p == ';' ? ++p : p)
+
+struct tagstack_item {
+ const char *tagname_begin;
+ const char *tagname_end;
+ const char *contents_begin;
+ struct tagstack_item *prev;
+ struct tagstack_item *next;
+};
+
+static struct tagstack_item *
+tagstack_push (struct tagstack_item **head, struct tagstack_item **tail)
+{
+ struct tagstack_item *ts = xmalloc(sizeof(struct tagstack_item));
+ if (*head == NULL)
+ {
+ *head = *tail = ts;
+ ts->prev = ts->next = NULL;
+ }
+ else
+ {
+ (*tail)->next = ts;
+ ts->prev = *tail;
+ *tail = ts;
+ ts->next = NULL;
+ }
+
+ return ts;
+}
+
+/* remove ts and everything after it from the stack */
+static void
+tagstack_pop (struct tagstack_item **head, struct tagstack_item **tail,
+ struct tagstack_item *ts)
+{
+ if (*head == NULL)
+ return;
+
+ if (ts == *tail)
+ {
+ if (ts == *head)
+ {
+ xfree (ts);
+ *head = *tail = NULL;
+ }
+ else
+ {
+ ts->prev->next = NULL;
+ *tail = ts->prev;
+ xfree (ts);
+ }
+ }
+ else
+ {
+ if (ts == *head)
+ {
+ *head = NULL;
+ }
+ *tail = ts->prev;
+
+ if (ts->prev)
+ {
+ ts->prev->next = NULL;
+ }
+ while (ts)
+ {
+ struct tagstack_item *p = ts->next;
+ xfree (ts);
+ ts = p;
+ }
+ }
+}
+
+static struct tagstack_item *
+tagstack_find (struct tagstack_item *tail, const char *tagname_begin,
+ const char *tagname_end)
+{
+ int len = tagname_end - tagname_begin;
+ while (tail)
+ {
+ if (len == (tail->tagname_end - tail->tagname_begin))
+ {
+ if (0 == strncasecmp (tail->tagname_begin, tagname_begin, len))
+ return tail;
+ }
+ tail = tail->prev;
+ }
+ return NULL;
+}
+
+/* Decode the HTML character entity at *PTR, considering END to be end
+ of buffer. It is assumed that the "&" character that marks the
+ beginning of the entity has been seen at *PTR-1. If a recognized
+ ASCII entity is seen, it is returned, and *PTR is moved to the end
+ of the entity. Otherwise, -1 is returned and *PTR left unmodified.
+
+ The recognized entities are: <, >, &, &apos, and ". */
+
+static int
+decode_entity (const char **ptr, const char *end)
+{
+ const char *p = *ptr;
+ int value = -1;
+
+ if (++p == end)
+ return -1;
+
+ switch (*p++)
+ {
+ case '#':
+ /* Process numeric entities "&#DDD;" and "&#xHH;". */
+ {
+ int digits = 0;
+ value = 0;
+ if (*p == 'x')
+ for (++p; value < 256 && p < end && c_isxdigit (*p); p++, digits++)
+ value = (value << 4) + XDIGIT_TO_NUM (*p);
+ else
+ for (; value < 256 && p < end && c_isdigit (*p); p++, digits++)
+ value = (value * 10) + (*p - '0');
+ if (!digits)
+ return -1;
+ /* Don't interpret 128+ codes and NUL because we cannot
+ portably reinserted them into HTML. */
+ if (!value || (value & ~0x7f))
+ return -1;
+ *ptr = SKIP_SEMI (p, 0);
+ return value;
+ }
+ /* Process named ASCII entities. */
+ case 'g':
+ if (ENT1 (p, 't'))
+ value = '>', *ptr = SKIP_SEMI (p, 1);
+ break;
+ case 'l':
+ if (ENT1 (p, 't'))
+ value = '<', *ptr = SKIP_SEMI (p, 1);
+ break;
+ case 'a':
+ if (ENT2 (p, 'm', 'p'))
+ value = '&', *ptr = SKIP_SEMI (p, 2);
+ else if (ENT3 (p, 'p', 'o', 's'))
+ /* handle &apos for the sake of the XML/XHTML crowd. */
+ value = '\'', *ptr = SKIP_SEMI (p, 3);
+ break;
+ case 'q':
+ if (ENT3 (p, 'u', 'o', 't'))
+ value = '\"', *ptr = SKIP_SEMI (p, 3);
+ break;
+ }
+ return value;
+}
+#undef ENT1
+#undef ENT2
+#undef ENT3
+#undef FITS
+#undef SKIP_SEMI
+
+enum {
+ AP_DOWNCASE = 1,
+ AP_DECODE_ENTITIES = 2,
+ AP_TRIM_BLANKS = 4
+};
/* Copy the text in the range [BEG, END) to POOL, optionally
performing operations specified by FLAGS. FLAGS may be any
- combination of AP_DOWNCASE, AP_PROCESS_ENTITIES and AP_TRIM_BLANKS
+ combination of AP_DOWNCASE, AP_DECODE_ENTITIES and AP_TRIM_BLANKS
with the following meaning:
* AP_DOWNCASE -- downcase all the letters;
- * AP_PROCESS_ENTITIES -- process the SGML entities and write out
- the decoded string. Recognized entities are <, >, &, ",
-   and the numerical entities.
+ * AP_DECODE_ENTITIES -- decode the named and numeric entities in
+ the ASCII range when copying the string.
* AP_TRIM_BLANKS -- ignore blanks at the beginning and at the end
- of text. */
+ of text, as well as embedded newlines. */
static void
convert_and_copy (struct pool *pool, const char *beg, const char *end, int flags)
{
int old_tail = pool->tail;
- int size;
- /* First, skip blanks if required. We must do this before entities
- are processed, so that blanks can still be inserted as, for
- instance, ` '. */
+ /* Skip blanks if required. We must do this before entities are
+ processed, so that blanks can still be inserted as, for instance,
+ ` '. */
if (flags & AP_TRIM_BLANKS)
{
- while (beg < end && ISSPACE (*beg))
- ++beg;
- while (end > beg && ISSPACE (end[-1]))
- --end;
+ while (beg < end && c_isspace (*beg))
+ ++beg;
+ while (end > beg && c_isspace (end[-1]))
+ --end;
}
- size = end - beg;
- if (flags & AP_PROCESS_ENTITIES)
+ if (flags & AP_DECODE_ENTITIES)
{
/* Grow the pool, then copy the text to the pool character by
- character, processing the encountered entities as we go
- along.
+ character, processing the encountered entities as we go
+ along.
- It's safe (and necessary) to grow the pool in advance because
- processing the entities can only *shorten* the string, it can
- never lengthen it. */
- POOL_GROW (pool, end - beg);
+ It's safe (and necessary) to grow the pool in advance because
+ processing the entities can only *shorten* the string, it can
+ never lengthen it. */
const char *from = beg;
- char *to = pool->contents + pool->tail;
+ char *to;
+ bool squash_newlines = !!(flags & AP_TRIM_BLANKS);
+
+ POOL_GROW (pool, end - beg);
+ to = pool->contents + pool->tail;
while (from < end)
- {
- if (*from != '&')
- *to++ = *from++;
- else
- {
- const char *save = from;
- int remain;
-
- if (++from == end)
- goto lose;
- remain = end - from;
-
- /* Process numeric entities "&#DDD;" and "&#xHH;". */
- if (*from == '#')
- {
- int numeric = 0, digits = 0;
- ++from;
- if (*from == 'x')
- {
- ++from;
- for (; from < end && ISXDIGIT (*from); from++, digits++)
- numeric = (numeric << 4) + XDIGIT_TO_NUM (*from);
- }
- else
- {
- for (; from < end && ISDIGIT (*from); from++, digits++)
- numeric = (numeric * 10) + (*from - '0');
- }
- if (!digits)
- goto lose;
- numeric &= 0xff;
- *to++ = numeric;
- }
-#define FROB(x) (remain >= (sizeof (x) - 1) \
- && 0 == memcmp (from, x, sizeof (x) - 1) \
- && (*(from + sizeof (x) - 1) == ';' \
- || remain == sizeof (x) - 1 \
- || !ISALNUM (*(from + sizeof (x) - 1))))
- else if (FROB ("lt"))
- *to++ = '<', from += 2;
- else if (FROB ("gt"))
- *to++ = '>', from += 2;
- else if (FROB ("amp"))
- *to++ = '&', from += 3;
- else if (FROB ("quot"))
- *to++ = '\"', from += 4;
- /* We don't implement the proposed "Added Latin 1"
- entities (except for nbsp), because it is unnecessary
- in the context of Wget, and would require hashing to
- work efficiently. */
- else if (FROB ("nbsp"))
- *to++ = 160, from += 4;
- else
- goto lose;
-#undef FROB
- /* If the entity was followed by `;', we step over the
- `;'. Otherwise, it was followed by either a
- non-alphanumeric or EOB, in which case we do nothing. */
- if (from < end && *from == ';')
- ++from;
- continue;
-
- lose:
- /* This was not an entity after all. Back out. */
- from = save;
- *to++ = *from++;
- }
- }
+ {
+ if (*from == '&')
+ {
+ int entity = decode_entity (&from, end);
+ if (entity != -1)
+ *to++ = entity;
+ else
+ *to++ = *from++;
+ }
+ else if ((*from == '\n' || *from == '\r') && squash_newlines)
+ ++from;
+ else
+ *to++ = *from++;
+ }
/* Verify that we haven't exceeded the original size. (It
- shouldn't happen, hence the assert.) */
+ shouldn't happen, hence the assert.) */
assert (to - (pool->contents + pool->tail) <= end - beg);
/* Make POOL's tail point to the position following the string
- we've written. */
+ we've written. */
pool->tail = to - pool->contents;
POOL_APPEND_CHR (pool, '\0');
}
{
char *p = pool->contents + old_tail;
for (; *p; p++)
- *p = TOLOWER (*p);
+ *p = c_tolower (*p);
}
}
\f
-/* Check whether the contents of [POS, POS+LENGTH) match any of the
- strings in the ARRAY. */
-static int
-array_allowed (const char **array, const char *beg, const char *end)
-{
- int length = end - beg;
- if (array)
- {
- for (; *array; array++)
- if (length >= strlen (*array)
- && !strncasecmp (*array, beg, length))
- break;
- if (!*array)
- return 0;
- }
- return 1;
-}
-\f
/* Originally we used to adhere to rfc 1866 here, and allowed only
letters, digits, periods, and hyphens as names (of tags or
attributes). However, this broke too many pages which used
* whitespace
* 8-bit and control chars
* characters that clearly cannot be part of name:
- '=', '>', '/'.
+ '=', '<', '>', '/'.
This only affects attribute and tag names; attribute values allow
an even greater variety of characters. */
-#define NAME_CHAR_P(x) ((x) > 32 && (x) < 127 \
- && (x) != '=' && (x) != '>' && (x) != '/')
+#define NAME_CHAR_P(x) ((x) > 32 && (x) < 127 \
+ && (x) != '=' && (x) != '<' && (x) != '>' \
+ && (x) != '/')
#ifdef STANDALONE
static int comment_backout_count;
advance_declaration (const char *beg, const char *end)
{
const char *p = beg;
- char quote_char = '\0'; /* shut up, gcc! */
+ char quote_char = '\0'; /* shut up, gcc! */
char ch;
enum {
AC_S_DASH4,
AC_S_QUOTE1,
AC_S_IN_QUOTE,
- AC_S_QUOTE2,
+ AC_S_QUOTE2
} state = AC_S_BANG;
if (beg == end)
while (state != AC_S_DONE && state != AC_S_BACKOUT)
{
if (p == end)
- state = AC_S_BACKOUT;
+ state = AC_S_BACKOUT;
switch (state)
- {
- case AC_S_DONE:
- case AC_S_BACKOUT:
- break;
- case AC_S_BANG:
- if (ch == '!')
- {
- ch = *p++;
- state = AC_S_DEFAULT;
- }
- else
- state = AC_S_BACKOUT;
- break;
- case AC_S_DEFAULT:
- switch (ch)
- {
- case '-':
- state = AC_S_DASH1;
- break;
- case ' ':
- case '\t':
- case '\r':
- case '\n':
- ch = *p++;
- break;
- case '>':
- state = AC_S_DONE;
- break;
- case '\'':
- case '\"':
- state = AC_S_QUOTE1;
- break;
- default:
- if (NAME_CHAR_P (ch))
- state = AC_S_DCLNAME;
- else
- state = AC_S_BACKOUT;
- break;
- }
- break;
- case AC_S_DCLNAME:
- if (ch == '-')
- state = AC_S_DASH1;
- else if (NAME_CHAR_P (ch))
- ch = *p++;
- else
- state = AC_S_DEFAULT;
- break;
- case AC_S_QUOTE1:
- /* We must use 0x22 because broken assert macros choke on
- '"' and '\"'. */
- assert (ch == '\'' || ch == 0x22);
- quote_char = ch; /* cheating -- I really don't feel like
- introducing more different states for
- different quote characters. */
- ch = *p++;
- state = AC_S_IN_QUOTE;
- break;
- case AC_S_IN_QUOTE:
- if (ch == quote_char)
- state = AC_S_QUOTE2;
- else
- ch = *p++;
- break;
- case AC_S_QUOTE2:
- assert (ch == quote_char);
- ch = *p++;
- state = AC_S_DEFAULT;
- break;
- case AC_S_DASH1:
- assert (ch == '-');
- ch = *p++;
- state = AC_S_DASH2;
- break;
- case AC_S_DASH2:
- switch (ch)
- {
- case '-':
- ch = *p++;
- state = AC_S_COMMENT;
- break;
- default:
- state = AC_S_BACKOUT;
- }
- break;
- case AC_S_COMMENT:
- switch (ch)
- {
- case '-':
- state = AC_S_DASH3;
- break;
- default:
- ch = *p++;
- break;
- }
- break;
- case AC_S_DASH3:
- assert (ch == '-');
- ch = *p++;
- state = AC_S_DASH4;
- break;
- case AC_S_DASH4:
- switch (ch)
- {
- case '-':
- ch = *p++;
- state = AC_S_DEFAULT;
- break;
- default:
- state = AC_S_COMMENT;
- break;
- }
- break;
- }
+ {
+ case AC_S_DONE:
+ case AC_S_BACKOUT:
+ break;
+ case AC_S_BANG:
+ if (ch == '!')
+ {
+ ch = *p++;
+ state = AC_S_DEFAULT;
+ }
+ else
+ state = AC_S_BACKOUT;
+ break;
+ case AC_S_DEFAULT:
+ switch (ch)
+ {
+ case '-':
+ state = AC_S_DASH1;
+ break;
+ case ' ':
+ case '\t':
+ case '\r':
+ case '\n':
+ ch = *p++;
+ break;
+ case '<':
+ case '>':
+ state = AC_S_DONE;
+ break;
+ case '\'':
+ case '\"':
+ state = AC_S_QUOTE1;
+ break;
+ default:
+ if (NAME_CHAR_P (ch))
+ state = AC_S_DCLNAME;
+ else
+ state = AC_S_BACKOUT;
+ break;
+ }
+ break;
+ case AC_S_DCLNAME:
+ if (ch == '-')
+ state = AC_S_DASH1;
+ else if (NAME_CHAR_P (ch))
+ ch = *p++;
+ else
+ state = AC_S_DEFAULT;
+ break;
+ case AC_S_QUOTE1:
+ /* We must use 0x22 because broken assert macros choke on
+ '"' and '\"'. */
+ assert (ch == '\'' || ch == 0x22);
+ quote_char = ch; /* cheating -- I really don't feel like
+ introducing more different states for
+ different quote characters. */
+ ch = *p++;
+ state = AC_S_IN_QUOTE;
+ break;
+ case AC_S_IN_QUOTE:
+ if (ch == quote_char)
+ state = AC_S_QUOTE2;
+ else
+ ch = *p++;
+ break;
+ case AC_S_QUOTE2:
+ assert (ch == quote_char);
+ ch = *p++;
+ state = AC_S_DEFAULT;
+ break;
+ case AC_S_DASH1:
+ assert (ch == '-');
+ ch = *p++;
+ state = AC_S_DASH2;
+ break;
+ case AC_S_DASH2:
+ switch (ch)
+ {
+ case '-':
+ ch = *p++;
+ state = AC_S_COMMENT;
+ break;
+ default:
+ state = AC_S_BACKOUT;
+ }
+ break;
+ case AC_S_COMMENT:
+ switch (ch)
+ {
+ case '-':
+ state = AC_S_DASH3;
+ break;
+ default:
+ ch = *p++;
+ break;
+ }
+ break;
+ case AC_S_DASH3:
+ assert (ch == '-');
+ ch = *p++;
+ state = AC_S_DASH4;
+ break;
+ case AC_S_DASH4:
+ switch (ch)
+ {
+ case '-':
+ ch = *p++;
+ state = AC_S_DEFAULT;
+ break;
+ default:
+ state = AC_S_COMMENT;
+ break;
+ }
+ break;
+ }
}
if (state == AC_S_BACKOUT)
switch (p[0])
{
case '>':
- if (p[-1] == '-' && p[-2] == '-')
- return p + 1;
- break;
+ if (p[-1] == '-' && p[-2] == '-')
+ return p + 1;
+ break;
case '-':
at_dash:
- if (p[-1] == '-')
- {
- at_dash_dash:
- if (++p == end) return NULL;
- switch (p[0])
- {
- case '>': return p + 1;
- case '-': goto at_dash_dash;
- }
- }
- else
- {
- if ((p += 2) >= end) return NULL;
- switch (p[0])
- {
- case '>':
- if (p[-1] == '-')
- return p + 1;
- break;
- case '-':
- goto at_dash;
- }
- }
+ if (p[-1] == '-')
+ {
+ at_dash_dash:
+ if (++p == end) return NULL;
+ switch (p[0])
+ {
+ case '>': return p + 1;
+ case '-': goto at_dash_dash;
+ }
+ }
+ else
+ {
+ if ((p += 2) >= end) return NULL;
+ switch (p[0])
+ {
+ case '>':
+ if (p[-1] == '-')
+ return p + 1;
+ break;
+ case '-':
+ goto at_dash;
+ }
+ }
}
return NULL;
}
\f
+/* Return true if the string containing of characters inside [b, e) is
+ present in hash table HT. */
+
+static bool
+name_allowed (const struct hash_table *ht, const char *b, const char *e)
+{
+ char *copy;
+ if (!ht)
+ return true;
+ BOUNDED_TO_ALLOCA (b, e, copy);
+ return hash_table_get (ht, copy) != NULL;
+}
+
/* Advance P (a char pointer), with the explicit intent of being able
to read the next character. If this is not possible, go to finish. */
-#define ADVANCE(p) do { \
- ++p; \
- if (p >= end) \
- goto finish; \
+#define ADVANCE(p) do { \
+ ++p; \
+ if (p >= end) \
+ goto finish; \
} while (0)
/* Skip whitespace, if any. */
-#define SKIP_WS(p) do { \
- while (ISSPACE (*p)) { \
- ADVANCE (p); \
- } \
+#define SKIP_WS(p) do { \
+ while (c_isspace (*p)) { \
+ ADVANCE (p); \
+ } \
} while (0)
/* Skip non-whitespace, if any. */
-#define SKIP_NON_WS(p) do { \
- while (!ISSPACE (*p)) { \
- ADVANCE (p); \
- } \
+#define SKIP_NON_WS(p) do { \
+ while (!c_isspace (*p)) { \
+ ADVANCE (p); \
+ } \
} while (0)
#ifdef STANDALONE
/* Map MAPFUN over HTML tags in TEXT, which is SIZE characters long.
MAPFUN will be called with two arguments: pointer to an initialized
- struct taginfo, and CLOSURE.
+ struct taginfo, and MAPARG.
- ALLOWED_TAG_NAMES should be a NULL-terminated array of tag names to
- be processed by this function. If it is NULL, all the tags are
- allowed. The same goes for attributes and ALLOWED_ATTRIBUTE_NAMES.
+ ALLOWED_TAGS and ALLOWED_ATTRIBUTES are hash tables the keys of
+ which are the tags and attribute names that this function should
+ use. If ALLOWED_TAGS is NULL, all tags are processed; if
+ ALLOWED_ATTRIBUTES is NULL, all attributes are returned.
(Obviously, the caller can filter out unwanted tags and attributes
just as well, but this is just an optimization designed to avoid
- unnecessary copying for tags/attributes which the caller doesn't
- want to know about. These lists are searched linearly; therefore,
- if you're interested in a large number of tags or attributes, you'd
- better set these to NULL and filter them out yourself with a
- hashing process most appropriate for your application.) */
+ unnecessary copying of tags/attributes which the caller doesn't
+ care about.) */
void
map_html_tags (const char *text, int size,
- const char **allowed_tag_names,
- const char **allowed_attribute_names,
- void (*mapfun) (struct taginfo *, void *),
- void *closure)
+ void (*mapfun) (struct taginfo *, void *), void *maparg,
+ int flags,
+ const struct hash_table *allowed_tags,
+ const struct hash_table *allowed_attributes)
{
/* storage for strings passed to MAPFUN callback; if 256 bytes is
too little, POOL_APPEND allocates more with malloc. */
struct attr_pair attr_pair_initial_storage[8];
int attr_pair_size = countof (attr_pair_initial_storage);
- int attr_pair_resized = 0;
+ bool attr_pair_resized = false;
struct attr_pair *pairs = attr_pair_initial_storage;
+ struct tagstack_item *head = NULL;
+ struct tagstack_item *tail = NULL;
+
if (!size)
return;
int nattrs, end_tag;
const char *tag_name_begin, *tag_name_end;
const char *tag_start_position;
- int uninteresting_tag;
+ bool uninteresting_tag;
look_for_tag:
POOL_REWIND (&pool);
declaration). */
if (*p == '!')
{
- if (!opt.strict_comments
- && p < end + 3 && p[1] == '-' && p[2] == '-')
- {
- /* If strict comments are not enforced and if we know
- we're looking at a comment, simply look for the
- terminating "-->". Non-strict is the default because
- it works in other browsers and most HTML writers can't
- be bothered with getting the comments right. */
- const char *comment_end = find_comment_end (p + 3, end);
- if (comment_end)
- p = comment_end;
- }
- else
- {
- /* Either in strict comment mode or looking at a non-empty
- declaration. Real declarations are much less likely to
- be misused the way comments are, so advance over them
- properly regardless of strictness. */
- p = advance_declaration (p, end);
- }
- if (p == end)
- goto finish;
- goto look_for_tag;
+ if (!(flags & MHT_STRICT_COMMENTS)
+ && p < end + 3 && p[1] == '-' && p[2] == '-')
+ {
+ /* If strict comments are not enforced and if we know
+ we're looking at a comment, simply look for the
+ terminating "-->". Non-strict is the default because
+ it works in other browsers and most HTML writers can't
+ be bothered with getting the comments right. */
+ const char *comment_end = find_comment_end (p + 3, end);
+ if (comment_end)
+ p = comment_end;
+ }
+ else
+ {
+ /* Either in strict comment mode or looking at a non-empty
+ declaration. Real declarations are much less likely to
+ be misused the way comments are, so advance over them
+ properly regardless of strictness. */
+ p = advance_declaration (p, end);
+ }
+ if (p == end)
+ goto finish;
+ goto look_for_tag;
}
else if (*p == '/')
{
- end_tag = 1;
- ADVANCE (p);
+ end_tag = 1;
+ ADVANCE (p);
}
tag_name_begin = p;
while (NAME_CHAR_P (*p))
goto look_for_tag;
tag_name_end = p;
SKIP_WS (p);
- if (end_tag && *p != '>')
+
+ if (!end_tag)
+ {
+ struct tagstack_item *ts = tagstack_push (&head, &tail);
+ if (ts)
+ {
+ ts->tagname_begin = tag_name_begin;
+ ts->tagname_end = tag_name_end;
+ ts->contents_begin = NULL;
+ }
+ }
+
+ if (end_tag && *p != '>' && *p != '<')
goto backout_tag;
- if (!array_allowed (allowed_tag_names, tag_name_begin, tag_name_end))
+ if (!name_allowed (allowed_tags, tag_name_begin, tag_name_end))
/* We can't just say "goto look_for_tag" here because we need
the loop below to properly advance over the tag's attributes. */
- uninteresting_tag = 1;
+ uninteresting_tag = true;
else
{
- uninteresting_tag = 0;
- convert_and_copy (&pool, tag_name_begin, tag_name_end, AP_DOWNCASE);
+ uninteresting_tag = false;
+ convert_and_copy (&pool, tag_name_begin, tag_name_end, AP_DOWNCASE);
}
/* Find the attributes. */
while (1)
{
- const char *attr_name_begin, *attr_name_end;
- const char *attr_value_begin, *attr_value_end;
- const char *attr_raw_value_begin, *attr_raw_value_end;
- int operation = AP_DOWNCASE; /* stupid compiler. */
-
- SKIP_WS (p);
-
- if (*p == '/')
- {
- /* A slash at this point means the tag is about to be
- closed. This is legal in XML and has been popularized
- in HTML via XHTML. */
- /* <foo a=b c=d /> */
- /* ^ */
- ADVANCE (p);
- SKIP_WS (p);
- if (*p != '>')
- goto backout_tag;
- }
-
- /* Check for end of tag definition. */
- if (*p == '>')
- break;
-
- /* Establish bounds of attribute name. */
- attr_name_begin = p; /* <foo bar ...> */
- /* ^ */
- while (NAME_CHAR_P (*p))
- ADVANCE (p);
- attr_name_end = p; /* <foo bar ...> */
- /* ^ */
- if (attr_name_begin == attr_name_end)
- goto backout_tag;
-
- /* Establish bounds of attribute value. */
- SKIP_WS (p);
- if (NAME_CHAR_P (*p) || *p == '/' || *p == '>')
- {
- /* Minimized attribute syntax allows `=' to be omitted.
+ const char *attr_name_begin, *attr_name_end;
+ const char *attr_value_begin, *attr_value_end;
+ const char *attr_raw_value_begin, *attr_raw_value_end;
+ int operation = AP_DOWNCASE; /* stupid compiler. */
+
+ SKIP_WS (p);
+
+ if (*p == '/')
+ {
+ /* A slash at this point means the tag is about to be
+ closed. This is legal in XML and has been popularized
+ in HTML via XHTML. */
+ /* <foo a=b c=d /> */
+ /* ^ */
+ ADVANCE (p);
+ SKIP_WS (p);
+ if (*p != '<' && *p != '>')
+ goto backout_tag;
+ }
+
+ /* Check for end of tag definition. */
+ if (*p == '<' || *p == '>')
+ break;
+
+ /* Establish bounds of attribute name. */
+ attr_name_begin = p; /* <foo bar ...> */
+ /* ^ */
+ while (NAME_CHAR_P (*p))
+ ADVANCE (p);
+ attr_name_end = p; /* <foo bar ...> */
+ /* ^ */
+ if (attr_name_begin == attr_name_end)
+ goto backout_tag;
+
+ /* Establish bounds of attribute value. */
+ SKIP_WS (p);
+
+ if (NAME_CHAR_P (*p) || *p == '/' || *p == '<' || *p == '>')
+ {
+ /* Minimized attribute syntax allows `=' to be omitted.
For example, <UL COMPACT> is a valid shorthand for <UL
COMPACT="compact">. Even if such attributes are not
useful to Wget, we need to support them, so that the
tags containing them can be parsed correctly. */
- attr_raw_value_begin = attr_value_begin = attr_name_begin;
- attr_raw_value_end = attr_value_end = attr_name_end;
- }
- else if (*p == '=')
- {
- ADVANCE (p);
- SKIP_WS (p);
- if (*p == '\"' || *p == '\'')
- {
- int newline_seen = 0;
- char quote_char = *p;
- attr_raw_value_begin = p;
- ADVANCE (p);
- attr_value_begin = p; /* <foo bar="baz"> */
- /* ^ */
- while (*p != quote_char)
- {
- if (!newline_seen && *p == '\n')
- {
- /* If a newline is seen within the quotes, it
- is most likely that someone forgot to close
- the quote. In that case, we back out to
- the value beginning, and terminate the tag
- at either `>' or the delimiter, whichever
- comes first. Such a tag terminated at `>'
- is discarded. */
- p = attr_value_begin;
- newline_seen = 1;
- continue;
- }
- else if (newline_seen && *p == '>')
- break;
- ADVANCE (p);
- }
- attr_value_end = p; /* <foo bar="baz"> */
- /* ^ */
- if (*p == quote_char)
- ADVANCE (p);
- else
- goto look_for_tag;
- attr_raw_value_end = p; /* <foo bar="baz"> */
- /* ^ */
- /* The AP_TRIM_BLANKS is there for buggy HTML
- generators that generate <a href=" foo"> instead of
- <a href="foo"> (Netscape ignores spaces as well.)
- If you really mean space, use &32; or %20. */
- operation = AP_PROCESS_ENTITIES | AP_TRIM_BLANKS;
- }
- else
- {
- attr_value_begin = p; /* <foo bar=baz> */
- /* ^ */
- /* According to SGML, a name token should consist only
- of alphanumerics, . and -. However, this is often
- violated by, for instance, `%' in `width=75%'.
- We'll be liberal and allow just about anything as
- an attribute value. */
- while (!ISSPACE (*p) && *p != '>')
- ADVANCE (p);
- attr_value_end = p; /* <foo bar=baz qux=quix> */
- /* ^ */
- if (attr_value_begin == attr_value_end)
- /* <foo bar=> */
- /* ^ */
- goto backout_tag;
- attr_raw_value_begin = attr_value_begin;
- attr_raw_value_end = attr_value_end;
- operation = AP_PROCESS_ENTITIES;
- }
- }
- else
- {
- /* We skipped the whitespace and found something that is
- neither `=' nor the beginning of the next attribute's
- name. Back out. */
- goto backout_tag; /* <foo bar [... */
- /* ^ */
- }
-
- /* If we're not interested in the tag, don't bother with any
+ attr_raw_value_begin = attr_value_begin = attr_name_begin;
+ attr_raw_value_end = attr_value_end = attr_name_end;
+ }
+ else if (*p == '=')
+ {
+ ADVANCE (p);
+ SKIP_WS (p);
+ if (*p == '\"' || *p == '\'')
+ {
+ bool newline_seen = false;
+ char quote_char = *p;
+ attr_raw_value_begin = p;
+ ADVANCE (p);
+ attr_value_begin = p; /* <foo bar="baz"> */
+ /* ^ */
+ while (*p != quote_char)
+ {
+ if (!newline_seen && *p == '\n')
+ {
+ /* If a newline is seen within the quotes, it
+ is most likely that someone forgot to close
+ the quote. In that case, we back out to
+ the value beginning, and terminate the tag
+ at either `>' or the delimiter, whichever
+ comes first. Such a tag terminated at `>'
+ is discarded. */
+ p = attr_value_begin;
+ newline_seen = true;
+ continue;
+ }
+ else if (newline_seen && (*p == '<' || *p == '>'))
+ break;
+ ADVANCE (p);
+ }
+ attr_value_end = p; /* <foo bar="baz"> */
+ /* ^ */
+ if (*p == quote_char)
+ ADVANCE (p);
+ else
+ goto look_for_tag;
+ attr_raw_value_end = p; /* <foo bar="baz"> */
+ /* ^ */
+ operation = AP_DECODE_ENTITIES;
+ if (flags & MHT_TRIM_VALUES)
+ operation |= AP_TRIM_BLANKS;
+ }
+ else
+ {
+ attr_value_begin = p; /* <foo bar=baz> */
+ /* ^ */
+ /* According to SGML, a name token should consist only
+ of alphanumerics, . and -. However, this is often
+ violated by, for instance, `%' in `width=75%'.
+ We'll be liberal and allow just about anything as
+ an attribute value. */
+ while (!c_isspace (*p) && *p != '<' && *p != '>')
+ ADVANCE (p);
+ attr_value_end = p; /* <foo bar=baz qux=quix> */
+ /* ^ */
+ if (attr_value_begin == attr_value_end)
+ /* <foo bar=> */
+ /* ^ */
+ goto backout_tag;
+ attr_raw_value_begin = attr_value_begin;
+ attr_raw_value_end = attr_value_end;
+ operation = AP_DECODE_ENTITIES;
+ }
+ }
+ else
+ {
+ /* We skipped the whitespace and found something that is
+ neither `=' nor the beginning of the next attribute's
+ name. Back out. */
+ goto backout_tag; /* <foo bar [... */
+ /* ^ */
+ }
+
+ /* If we're not interested in the tag, don't bother with any
of the attributes. */
- if (uninteresting_tag)
- continue;
+ if (uninteresting_tag)
+ continue;
- /* If we aren't interested in the attribute, skip it. We
+ /* If we aren't interested in the attribute, skip it. We
cannot do this test any sooner, because our text pointer
needs to correctly advance over the attribute. */
- if (allowed_attribute_names
- && !array_allowed (allowed_attribute_names, attr_name_begin,
- attr_name_end))
- continue;
-
- GROW_ARRAY (pairs, attr_pair_size, nattrs + 1, attr_pair_resized,
- struct attr_pair);
-
- pairs[nattrs].name_pool_index = pool.tail;
- convert_and_copy (&pool, attr_name_begin, attr_name_end, AP_DOWNCASE);
-
- pairs[nattrs].value_pool_index = pool.tail;
- convert_and_copy (&pool, attr_value_begin, attr_value_end, operation);
- pairs[nattrs].value_raw_beginning = attr_raw_value_begin;
- pairs[nattrs].value_raw_size = (attr_raw_value_end
- - attr_raw_value_begin);
- ++nattrs;
+ if (!name_allowed (allowed_attributes, attr_name_begin, attr_name_end))
+ continue;
+
+ GROW_ARRAY (pairs, attr_pair_size, nattrs + 1, attr_pair_resized,
+ struct attr_pair);
+
+ pairs[nattrs].name_pool_index = pool.tail;
+ convert_and_copy (&pool, attr_name_begin, attr_name_end, AP_DOWNCASE);
+
+ pairs[nattrs].value_pool_index = pool.tail;
+ convert_and_copy (&pool, attr_value_begin, attr_value_end, operation);
+ pairs[nattrs].value_raw_beginning = attr_raw_value_begin;
+ pairs[nattrs].value_raw_size = (attr_raw_value_end
+ - attr_raw_value_begin);
+ ++nattrs;
+ }
+
+ if (!end_tag && tail && (tail->tagname_begin == tag_name_begin))
+ {
+ tail->contents_begin = p+1;
}
if (uninteresting_tag)
{
- ADVANCE (p);
- goto look_for_tag;
+ ADVANCE (p);
+ goto look_for_tag;
}
/* By now, we have a valid tag with a name and zero or more
{
int i;
struct taginfo taginfo;
+ struct tagstack_item *ts = NULL;
taginfo.name = pool.contents;
taginfo.end_tag_p = end_tag;
taginfo.nattrs = nattrs;
/* We fill in the char pointers only now, when pool can no
- longer get realloc'ed. If we did that above, we could get
- hosed by reallocation. Obviously, after this point, the pool
- may no longer be grown. */
+ longer get realloc'ed. If we did that above, we could get
+ hosed by reallocation. Obviously, after this point, the pool
+ may no longer be grown. */
for (i = 0; i < nattrs; i++)
- {
- pairs[i].name = pool.contents + pairs[i].name_pool_index;
- pairs[i].value = pool.contents + pairs[i].value_pool_index;
- }
+ {
+ pairs[i].name = pool.contents + pairs[i].name_pool_index;
+ pairs[i].value = pool.contents + pairs[i].value_pool_index;
+ }
taginfo.attrs = pairs;
taginfo.start_position = tag_start_position;
taginfo.end_position = p + 1;
- /* Ta-dam! */
- (*mapfun) (&taginfo, closure);
- ADVANCE (p);
+ taginfo.contents_begin = NULL;
+ taginfo.contents_end = NULL;
+
+ if (end_tag)
+ {
+ ts = tagstack_find (tail, tag_name_begin, tag_name_end);
+ if (ts)
+ {
+ if (ts->contents_begin)
+ {
+ taginfo.contents_begin = ts->contents_begin;
+ taginfo.contents_end = tag_start_position;
+ }
+ tagstack_pop (&head, &tail, ts);
+ }
+ }
+
+ mapfun (&taginfo, maparg);
+ if (*p != '<')
+ ADVANCE (p);
}
goto look_for_tag;
POOL_FREE (&pool);
if (attr_pair_resized)
xfree (pairs);
+ /* pop any tag stack that's left */
+ tagstack_pop (&head, &tail, head);
}
#undef ADVANCE
int main ()
{
int size = 256;
- char *x = (char *)xmalloc (size);
+ char *x = xmalloc (size);
int length = 0;
int read_count;
int tag_counter = 0;
{
length += read_count;
size <<= 1;
- x = (char *)xrealloc (x, size);
+ x = xrealloc (x, size);
}
- map_html_tags (x, length, NULL, NULL, test_mapper, &tag_counter);
+ map_html_tags (x, length, test_mapper, &tag_counter, 0, NULL, NULL);
printf ("TAGS: %d\n", tag_counter);
printf ("Tag backouts: %d\n", tag_backout_count);
printf ("Comment backouts: %d\n", comment_backout_count);
projects / wget / blobdiff