diff options
author | John Mark Bell <jmb@netsurf-browser.org> | 2008-07-31 14:43:32 +0000 |
---|---|---|
committer | John Mark Bell <jmb@netsurf-browser.org> | 2008-07-31 14:43:32 +0000 |
commit | 6d3950b141b1cdb1f29e929c8c8156562c7ccbc7 (patch) | |
tree | d2bb92960c4c3b54f083f1307241d5fdbac7fdfd /src/input/filter.c | |
parent | 2f7a0e956c151c900789a02d8f850bfe79b54071 (diff) | |
download | libhubbub-6d3950b141b1cdb1f29e929c8c8156562c7ccbc7.tar.gz libhubbub-6d3950b141b1cdb1f29e929c8c8156562c7ccbc7.tar.bz2 |
Merged revisions 4631-4838 via svnmerge from
svn://source.netsurf-browser.org/branches/takkaria/hubbub-parserutils
........
r4631 | takkaria | 2008-07-13 12:54:30 +0100 (Sun, 13 Jul 2008) | 2 lines
Initial hatchet job moving to libparserutils (search and replace and a bit of cleaning up). This doesn't compile.
........
r4632 | takkaria | 2008-07-13 15:28:52 +0100 (Sun, 13 Jul 2008) | 2 lines
libparserutilize everything up to the "before attribute name" state. (Not compiling)
........
r4633 | takkaria | 2008-07-13 15:32:14 +0100 (Sun, 13 Jul 2008) | 2 lines
Replace all uses of "current_{comment|chars}" with just "chars".
........
r4634 | takkaria | 2008-07-13 16:12:06 +0100 (Sun, 13 Jul 2008) | 2 lines
Fix lots of compile errors, lpuise "before attribute name" state.
........
r4636 | takkaria | 2008-07-13 17:23:17 +0100 (Sun, 13 Jul 2008) | 2 lines
Finish lpuising the tag states, apart from character references.
........
r4637 | takkaria | 2008-07-13 19:58:52 +0100 (Sun, 13 Jul 2008) | 2 lines
lpuise the comment states.
........
r4638 | takkaria | 2008-07-13 20:04:31 +0100 (Sun, 13 Jul 2008) | 2 lines
Switch to setting hubbub_string::len to 0 instead of hubbub_string::ptr to NULL to indicate an empty buffer, as it was previously.
........
r4639 | takkaria | 2008-07-13 21:02:11 +0100 (Sun, 13 Jul 2008) | 2 lines
"lpu up" about half of the DOCTYPE handling stages.
........
r4640 | takkaria | 2008-07-13 21:23:00 +0100 (Sun, 13 Jul 2008) | 2 lines
Finish off LPUing the doctype modes.
........
r4641 | takkaria | 2008-07-13 21:37:33 +0100 (Sun, 13 Jul 2008) | 2 lines
The tokeniser uses lpu apart from the entity matcher, now.
........
r4643 | takkaria | 2008-07-14 01:20:36 +0100 (Mon, 14 Jul 2008) | 2 lines
Fix up the character reference matching stuff--still not properly dealt with, but compiles futher.
........
r4644 | takkaria | 2008-07-14 01:24:49 +0100 (Mon, 14 Jul 2008) | 2 lines
Get the tokeniser compiling in its LPU'd form.
........
r4645 | takkaria | 2008-07-14 01:26:34 +0100 (Mon, 14 Jul 2008) | 2 lines
Remember to advance the stream position after emitting tokens.
........
r4646 | takkaria | 2008-07-14 01:34:36 +0100 (Mon, 14 Jul 2008) | 2 lines
Nuke the src/input directory and start work on the treebuilder.
........
r4647 | takkaria | 2008-07-14 01:56:27 +0100 (Mon, 14 Jul 2008) | 2 lines
Get hubbub building in its LPU'd form.
........
r4648 | takkaria | 2008-07-14 02:41:03 +0100 (Mon, 14 Jul 2008) | 2 lines
Get the tokeniser2 testrunner working.
........
r4649 | takkaria | 2008-07-14 02:48:55 +0100 (Mon, 14 Jul 2008) | 2 lines
Fix test LDFLAGS so things link properly.
........
r4650 | takkaria | 2008-07-14 16:25:51 +0100 (Mon, 14 Jul 2008) | 2 lines
Get testcases compiling, remove ones now covered by libparserutils.
........
r4651 | takkaria | 2008-07-14 16:37:09 +0100 (Mon, 14 Jul 2008) | 2 lines
Remove more tests covered by libpu.
........
r4652 | takkaria | 2008-07-14 17:53:18 +0100 (Mon, 14 Jul 2008) | 2 lines
Fix up the tokeniser a bit.
........
r4653 | takkaria | 2008-07-14 19:02:15 +0100 (Mon, 14 Jul 2008) | 3 lines
- Remove the buffer_handler stuff from hubbub
- Add the basics of a buffer for attribute values and text.
........
r4654 | takkaria | 2008-07-14 20:00:45 +0100 (Mon, 14 Jul 2008) | 2 lines
Get character references working in attribute values, start trying to make them work in character tokens.
........
r4656 | takkaria | 2008-07-14 23:28:52 +0100 (Mon, 14 Jul 2008) | 2 lines
Get entities working a bit better.
........
r4657 | takkaria | 2008-07-14 23:37:16 +0100 (Mon, 14 Jul 2008) | 2 lines
Get entities working properly. (!)
........
r4658 | takkaria | 2008-07-14 23:56:10 +0100 (Mon, 14 Jul 2008) | 2 lines
Make doctypes work a bit better.
........
r4659 | takkaria | 2008-07-15 00:18:49 +0100 (Tue, 15 Jul 2008) | 2 lines
Get DOCTYPEs working.
........
r4660 | takkaria | 2008-07-15 00:26:36 +0100 (Tue, 15 Jul 2008) | 2 lines
Fix CDATA sections.
........
r4661 | takkaria | 2008-07-15 01:01:16 +0100 (Tue, 15 Jul 2008) | 2 lines
Get comments working again.
........
r4662 | takkaria | 2008-07-15 01:14:19 +0100 (Tue, 15 Jul 2008) | 2 lines
Fix EOF in "after attribute name" state.
........
r4664 | takkaria | 2008-07-15 01:30:27 +0100 (Tue, 15 Jul 2008) | 2 lines
Put the tests in better order, remove one now superceded with libpu.
........
r4665 | takkaria | 2008-07-15 01:46:29 +0100 (Tue, 15 Jul 2008) | 2 lines
Remove a lot of now-redunant clearings of the current stream offset.
........
r4667 | jmb | 2008-07-15 11:56:54 +0100 (Tue, 15 Jul 2008) | 2 lines
Completely purge charset stuff from hubbub. Parserutils handles this now.
........
r4677 | takkaria | 2008-07-15 21:03:42 +0100 (Tue, 15 Jul 2008) | 2 lines
Get more tests passing, handle NUL bytes in data state.
........
r4694 | takkaria | 2008-07-18 17:55:44 +0100 (Fri, 18 Jul 2008) | 3 lines
- Handle CRs correctly in some token states.
- Handle NULs correctly in the CDATA state.
........
r4706 | takkaria | 2008-07-19 14:58:48 +0100 (Sat, 19 Jul 2008) | 2 lines
Improve the tokeniser2 output a bit.
........
r4721 | takkaria | 2008-07-21 20:57:29 +0100 (Mon, 21 Jul 2008) | 2 lines
Get a better framework in place to allow switching to using a buffer mid-collect. This fails a couple of testcases and doesn't implement proper CR or NUL support yet.
........
r4725 | takkaria | 2008-07-23 17:20:07 +0100 (Wed, 23 Jul 2008) | 2 lines
Make comment tokens in tokeniser2 display both expected and actual output.
........
r4726 | takkaria | 2008-07-23 19:10:23 +0100 (Wed, 23 Jul 2008) | 4 lines
- Add FINISH() macro which stops using buffered character collection.
- Make the encoding U+FFFD in UTF-8 a global varabile, for sanity
- Make the bogus comment state deal with NULs correctly.
........
r4730 | takkaria | 2008-07-24 00:35:16 +0100 (Thu, 24 Jul 2008) | 2 lines
Try to get NUL bytes handled as the spec says.
........
r4731 | takkaria | 2008-07-24 00:40:59 +0100 (Thu, 24 Jul 2008) | 2 lines
Get CRs working in the data state.
........
r4732 | takkaria | 2008-07-24 00:47:45 +0100 (Thu, 24 Jul 2008) | 2 lines
Set force-quirks correctly when failing to match PUBLIC or SYSTEM in DOCTYPEs.
........
r4773 | takkaria | 2008-07-28 15:34:41 +0100 (Mon, 28 Jul 2008) | 2 lines
Fix up the tokeniser, finally.
........
r4801 | takkaria | 2008-07-29 15:59:31 +0100 (Tue, 29 Jul 2008) | 2 lines
Refactor macros a bit.
........
r4802 | takkaria | 2008-07-29 16:04:17 +0100 (Tue, 29 Jul 2008) | 2 lines
Do s/HUBBUB_TOKENISER_STATE_/STATE_/, for shorter line lengths.
........
r4805 | takkaria | 2008-07-29 16:58:37 +0100 (Tue, 29 Jul 2008) | 4 lines
Start cleaning up the hubbub tokeniser;
- refactor to use new inline emit_character_token() and emit_current_tag() functions; makes code clearer
- check EOF before using the CHAR() macro, so eventually it can be removed.
........
r4806 | takkaria | 2008-07-29 17:45:36 +0100 (Tue, 29 Jul 2008) | 2 lines
More cleanup like the previous commit.
........
r4807 | takkaria | 2008-07-29 19:48:44 +0100 (Tue, 29 Jul 2008) | 2 lines
Rewrite comment-handling code to be just the one function, whilst updating it to handle CRs and NULs properly. (All comments now always use the buffer.)
........
r4820 | takkaria | 2008-07-30 14:14:49 +0100 (Wed, 30 Jul 2008) | 2 lines
Finish off the first sweep of cleaning up and refactoring the tokeniser.
........
r4821 | takkaria | 2008-07-30 15:12:22 +0100 (Wed, 30 Jul 2008) | 2 lines
Add copyright statement.
........
r4822 | takkaria | 2008-07-30 17:23:01 +0100 (Wed, 30 Jul 2008) | 2 lines
Apply changes made to tokeniser2 to tokeniser3.
........
r4829 | takkaria | 2008-07-31 01:59:07 +0100 (Thu, 31 Jul 2008) | 4 lines
- Make the tokeniser save everything into the buffer, at least for now.
- Fix logic errors introduced in refactoring
- Avoid emitting more tokens than we have to (e.g. instead of emitting "<>" and switching back to the data state, just switch back to the data state and let it take care of it)
........
r4830 | takkaria | 2008-07-31 02:03:08 +0100 (Thu, 31 Jul 2008) | 2 lines
Small treebuilder <isindex> fix.
........
r4831 | takkaria | 2008-07-31 02:32:29 +0100 (Thu, 31 Jul 2008) | 2 lines
Stop holding on to pointers to character data across treebuilder calls.
........
r4832 | takkaria | 2008-07-31 02:45:09 +0100 (Thu, 31 Jul 2008) | 18 lines
Merge revisions 4620-4831 from trunk hubbub to libinputstream hubbub, modulo one change to test/Makefile which makes the linker choke when linking tests.
------------------------------------------------------------------------
r4666 | jmb | 2008-07-15 11:52:13 +0100 (Tue, 15 Jul 2008) | 3 lines
Make tree2 perform reference counting.
Fix bits of the treebuilder to perform reference counting correctly in the face of *result not pointing to the same object as the node passed in to the treebuilder client callbacks.
------------------------------------------------------------------------
r4668 | jmb | 2008-07-15 12:37:30 +0100 (Tue, 15 Jul 2008) | 2 lines
Fully document treebuilder callbacks.
------------------------------------------------------------------------
r4675 | takkaria | 2008-07-15 21:01:03 +0100 (Tue, 15 Jul 2008) | 2 lines
Fix memory leak in tokeniser2.
------------------------------------------------------------------------
........
r4834 | jmb | 2008-07-31 09:57:51 +0100 (Thu, 31 Jul 2008) | 2 lines
Fix infinite loop in charset detector
........
r4835 | jmb | 2008-07-31 13:01:24 +0100 (Thu, 31 Jul 2008) | 2 lines
Actually store namespaces on formatting list. Otherwise we read uninitialised memory. Add some semblance of filling allocations with junk to myrealloc().
........
r4836 | jmb | 2008-07-31 13:06:07 +0100 (Thu, 31 Jul 2008) | 2 lines
Lose debug again
........
r4837 | jmb | 2008-07-31 15:09:19 +0100 (Thu, 31 Jul 2008) | 2 lines
Lose obsolete testdata (this is now part of lpu)
........
svn path=/trunk/hubbub/; revision=4839
Diffstat (limited to 'src/input/filter.c')
-rw-r--r-- | src/input/filter.c | 380 |
1 files changed, 0 insertions, 380 deletions
diff --git a/src/input/filter.c b/src/input/filter.c deleted file mode 100644 index 7a97840..0000000 --- a/src/input/filter.c +++ /dev/null @@ -1,380 +0,0 @@ -/* - * This file is part of Hubbub. - * Licensed under the MIT License, - * http://www.opensource.org/licenses/mit-license.php - * Copyright 2007 John-Mark Bell <jmb@netsurf-browser.org> - */ - -#include <errno.h> -#include <stdbool.h> -#include <stdlib.h> -#include <string.h> - -#include "charset/aliases.h" -#include "charset/codec.h" -#include "utils/utils.h" - -#include "input/filter.h" - - -/** Input filter */ -struct hubbub_filter { - hubbub_charsetcodec *read_codec; /**< Read codec */ - hubbub_charsetcodec *write_codec; /**< Write codec */ - - uint32_t filter_output[2]; /**< Filter output buffer */ - uint32_t last_filter_char; /**< Last filtered character */ - - uint32_t pivot_buf[64]; /**< Conversion pivot buffer */ - - bool leftover; /**< Data remains from last call */ - uint8_t *pivot_left; /**< Remaining pivot to write */ - size_t pivot_len; /**< Length of pivot remaining */ - - struct { - uint16_t encoding; /**< Input encoding */ - } settings; /**< Filter settings */ - - hubbub_alloc alloc; /**< Memory (de)allocation function */ - void *pw; /**< Client private data */ -}; - -static hubbub_error hubbub_filter_set_defaults(hubbub_filter *input); -static hubbub_error hubbub_filter_set_encoding(hubbub_filter *input, - const char *enc); -static hubbub_error read_character_filter(uint32_t c, - uint32_t **output, size_t *outputlen, void *pw); - -/** - * Create an input filter - * - * \param int_enc Desired encoding of document - * \param alloc Function used to (de)allocate data - * \param pw Pointer to client-specific private data (may be NULL) - * \return Pointer to filter instance, or NULL on failure - */ -hubbub_filter *hubbub_filter_create(const char *int_enc, - hubbub_alloc alloc, void *pw) -{ - hubbub_filter *filter; - - if (alloc == NULL) - return NULL; - - filter = alloc(NULL, sizeof(*filter), pw); - if (!filter) - return NULL; - - filter->last_filter_char = 0; - - filter->leftover = false; - filter->pivot_left = NULL; - filter->pivot_len = 0; - - filter->alloc = alloc; - filter->pw = pw; - - if (hubbub_filter_set_defaults(filter) != HUBBUB_OK) { - filter->alloc(filter, 0, pw); - return NULL; - } - - filter->write_codec = hubbub_charsetcodec_create(int_enc, alloc, pw); - if (filter->write_codec == NULL) { - if (filter->read_codec != NULL) - hubbub_charsetcodec_destroy(filter->read_codec); - filter->alloc(filter, 0, pw); - return NULL; - } - - return filter; -} - -/** - * Destroy an input filter - * - * \param input Pointer to filter instance - */ -void hubbub_filter_destroy(hubbub_filter *input) -{ - if (input == NULL) - return; - - if (input->read_codec != NULL) - hubbub_charsetcodec_destroy(input->read_codec); - - if (input->write_codec != NULL) - hubbub_charsetcodec_destroy(input->write_codec); - - input->alloc(input, 0, input->pw); - - return; -} - -/** - * Configure an input filter - * - * \param input Pointer to filter instance - * \param type Input option type to configure - * \param params Option-specific parameters - * \return HUBBUB_OK on success, appropriate error otherwise - */ -hubbub_error hubbub_filter_setopt(hubbub_filter *input, - hubbub_filter_opttype type, - hubbub_filter_optparams *params) -{ - hubbub_error error = HUBBUB_OK; - - if (input == NULL || params == NULL) - return HUBBUB_BADPARM; - - switch (type) { - case HUBBUB_FILTER_SET_ENCODING: - error = hubbub_filter_set_encoding(input, - params->encoding.name); - break; - } - - return error; -} - -/** - * Process a chunk of data - * - * \param input Pointer to filter instance - * \param data Pointer to pointer to input buffer - * \param len Pointer to length of input buffer - * \param output Pointer to pointer to output buffer - * \param outlen Pointer to length of output buffer - * \return HUBBUB_OK on success, appropriate error otherwise - * - * Call this with an input buffer length of 0 to flush any buffers. - */ -hubbub_error hubbub_filter_process_chunk(hubbub_filter *input, - const uint8_t **data, size_t *len, - uint8_t **output, size_t *outlen) -{ - hubbub_error read_error, write_error; - - if (input == NULL || data == NULL || *data == NULL || len == NULL || - output == NULL || *output == NULL || outlen == NULL) - return HUBBUB_BADPARM; - - if (input->leftover) { - /* Some data left to be written from last call */ - - /* Attempt to flush the remaining data. */ - write_error = hubbub_charsetcodec_encode(input->write_codec, - (const uint8_t **) &input->pivot_left, - &input->pivot_len, - output, outlen); - - if (write_error != HUBBUB_OK) { - return write_error; - } - - /* And clear leftover */ - input->pivot_left = NULL; - input->pivot_len = 0; - input->leftover = false; - } - - while (*len > 0) { - size_t pivot_len = sizeof(input->pivot_buf); - uint8_t *pivot = (uint8_t *) input->pivot_buf; - - read_error = hubbub_charsetcodec_decode(input->read_codec, - data, len, - (uint8_t **) &pivot, &pivot_len); - - pivot = (uint8_t *) input->pivot_buf; - pivot_len = sizeof(input->pivot_buf) - pivot_len; - - if (pivot_len > 0) { - write_error = hubbub_charsetcodec_encode( - input->write_codec, - (const uint8_t **) &pivot, - &pivot_len, - output, outlen); - - if (write_error != HUBBUB_OK) { - input->leftover = true; - input->pivot_left = pivot; - input->pivot_len = pivot_len; - - return write_error; - } - } - - if (read_error != HUBBUB_OK && read_error != HUBBUB_NOMEM) - return read_error; - } - - return HUBBUB_OK; -} - -/** - * Reset an input filter's state - * - * \param input The input filter to reset - * \param HUBBUB_OK on success, appropriate error otherwise - */ -hubbub_error hubbub_filter_reset(hubbub_filter *input) -{ - hubbub_error error; - - if (input == NULL) - return HUBBUB_BADPARM; - - /* Clear pivot buffer leftovers */ - input->pivot_left = NULL; - input->pivot_len = 0; - input->leftover = false; - - /* Reset read codec */ - error = hubbub_charsetcodec_reset(input->read_codec); - if (error != HUBBUB_OK) - return error; - - /* Reset write codec */ - error = hubbub_charsetcodec_reset(input->write_codec); - if (error != HUBBUB_OK) - return error; - - return HUBBUB_OK; -} - -/** - * Set an input filter's default settings - * - * \param input Input filter to configure - * \return HUBBUB_OK on success, appropriate error otherwise - */ -hubbub_error hubbub_filter_set_defaults(hubbub_filter *input) -{ - hubbub_error error; - - if (input == NULL) - return HUBBUB_BADPARM; - - input->read_codec = NULL; - input->write_codec = NULL; - input->settings.encoding = 0; - error = hubbub_filter_set_encoding(input, "ISO-8859-1"); - if (error != HUBBUB_OK) - return error; - - return HUBBUB_OK; -} - -/** - * Set an input filter's encoding - * - * \param input Input filter to configure - * \param enc Encoding name - * \return HUBBUB_OK on success, appropriate error otherwise - */ -hubbub_error hubbub_filter_set_encoding(hubbub_filter *input, - const char *enc) -{ - const char *old_enc; - uint16_t mibenum; - hubbub_error error; - hubbub_charsetcodec_optparams params; - - if (input == NULL || enc == NULL) - return HUBBUB_BADPARM; - - mibenum = hubbub_mibenum_from_name(enc, strlen(enc)); - if (mibenum == 0) - return HUBBUB_INVALID; - - /* Exit early if we're already using this encoding */ - if (input->settings.encoding == mibenum) - return HUBBUB_OK; - - old_enc = hubbub_mibenum_to_name(input->settings.encoding); - if (old_enc == NULL) - old_enc = "ISO-8859-1"; - - if (input->read_codec != NULL) - hubbub_charsetcodec_destroy(input->read_codec); - - input->read_codec = hubbub_charsetcodec_create(enc, input->alloc, - input->pw); - if (input->read_codec == NULL) - return HUBBUB_NOMEM; - - /* Register filter function */ - params.filter_func.filter = read_character_filter; - params.filter_func.pw = (void *) input; - error = hubbub_charsetcodec_setopt(input->read_codec, - HUBBUB_CHARSETCODEC_FILTER_FUNC, - (hubbub_charsetcodec_optparams *) ¶ms); - if (error != HUBBUB_OK) - return error; - - input->settings.encoding = mibenum; - - return HUBBUB_OK; -} - -/** - * Character filter function for read characters - * - * \param c The read character (UCS4 - host byte order) - * \param output Pointer to pointer to output buffer (filled on exit) - * \param outputlen Pointer to output buffer length (filled on exit) - * \param pw Pointer to client-specific private data. - * \return HUBBUB_OK on success, appropriate error otherwise - */ -hubbub_error read_character_filter(uint32_t c, uint32_t **output, - size_t *outputlen, void *pw) -{ - hubbub_filter *input = (hubbub_filter *) pw; - size_t len; - - if (output == NULL || outputlen == NULL || pw == NULL) - return HUBBUB_BADPARM; - - /* Line ending normalisation: - * CRLF -> LF (trap CR and let LF through unmodified) - * CR -> LF (trap CR and convert to LF if not CRLF) - * LF -> LF (leave LF alone) - */ - -#define NUL (0x00000000) -#define CR (0x0000000D) -#define LF (0x0000000A) -#define REP (0x0000FFFD) - - /* Replace NUL (U+0000) characters in input with U+FFFD */ - if (c == NUL) - c = REP; - - if (c == CR) { - /* Convert CRs to LFs straight away */ - input->filter_output[0] = LF; - len = 1; - } else if (input->last_filter_char == CR && c == LF) { - /* Trap this LF */ - len = 0; - } else { - /* Let character through unchanged */ - input->filter_output[0] = c; - len = 1; - } - - -#undef NUL -#undef CR -#undef LF -#undef REP - - input->last_filter_char = c; - - *output = input->filter_output; - *outputlen = len; - - return HUBBUB_OK; -} |