7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen#include "lib.h"

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen#include "buffer.h"

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen#include "str.h"

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen#include "istream.h"

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen#include "mail-html2text.h"

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen#include "message-parser.h"

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen#include "message-decoder.h"

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen#include "message-snippet.h"

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainenenum snippet_state {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen /* beginning of the line */

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen SNIPPET_STATE_NEWLINE = 0,

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen /* within normal text */

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen SNIPPET_STATE_NORMAL,

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen /* within quoted text - skip until EOL */

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen SNIPPET_STATE_QUOTED

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen};

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainenstruct snippet_context {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen string_t *snippet;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen unsigned int chars_left;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen enum snippet_state state;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen bool add_whitespace;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen struct mail_html2text *html2text;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen buffer_t *plain_output;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen};

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainenstatic bool snippet_generate(struct snippet_context *ctx,

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen const unsigned char *data, size_t size)

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen{

73083841b92505091b7d1070554930651b91b701Aki Tuomi size_t i, count;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (ctx->html2text != NULL) {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen buffer_set_used_size(ctx->plain_output, 0);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen mail_html2text_more(ctx->html2text, data, size,

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx->plain_output);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen data = ctx->plain_output->data;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen size = ctx->plain_output->used;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen }

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen /* message-decoder should feed us only valid and complete

73083841b92505091b7d1070554930651b91b701Aki Tuomi

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen for (i = 0; i < size; i += count) {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen count = 1;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen switch (ctx->state) {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen case SNIPPET_STATE_NEWLINE:

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (data[i] == '>' && ctx->html2text == NULL) {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx->state = SNIPPET_STATE_QUOTED;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen break;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen }

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx->state = SNIPPET_STATE_NORMAL;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen /* fallthrough */

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen case SNIPPET_STATE_NORMAL:

73083841b92505091b7d1070554930651b91b701Aki Tuomi if (size-i >= 3 &&

099737facc44553349a15521cdfcc65b112054caTimo Sirainen ((data[i] == 0xEF && data[i+1] == 0xBB && data[i+2] == 0xBF) ||

099737facc44553349a15521cdfcc65b112054caTimo Sirainen (data[i] == 0xBF && data[i+1] == 0xBB && data[i+2] == 0xEF))) {

73083841b92505091b7d1070554930651b91b701Aki Tuomi count += 2; /* because we skip +1 next */

73083841b92505091b7d1070554930651b91b701Aki Tuomi break;

73083841b92505091b7d1070554930651b91b701Aki Tuomi }

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (data[i] == '\r' || data[i] == '\n' ||

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen data[i] == '\t' || data[i] == ' ') {

73083841b92505091b7d1070554930651b91b701Aki Tuomi /* skip any leading whitespace */

73083841b92505091b7d1070554930651b91b701Aki Tuomi if (str_len(ctx->snippet) > 1)

73083841b92505091b7d1070554930651b91b701Aki Tuomi ctx->add_whitespace = TRUE;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (data[i] == '\n')

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx->state = SNIPPET_STATE_NEWLINE;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen break;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen }

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (ctx->add_whitespace) {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen str_append_c(ctx->snippet, ' ');

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx->add_whitespace = FALSE;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (ctx->chars_left-- == 0)

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen return FALSE;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen }

a05819736f348d0c5ac8b4966ac6b04c21e1a391Timo Sirainen if (ctx->chars_left == 0)

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen return FALSE;

a05819736f348d0c5ac8b4966ac6b04c21e1a391Timo Sirainen ctx->chars_left--;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen count = uni_utf8_char_bytes(data[i]);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen i_assert(i + count <= size);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen str_append_n(ctx->snippet, data + i, count);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen break;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen case SNIPPET_STATE_QUOTED:

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (data[i] == '\n')

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx->state = SNIPPET_STATE_NEWLINE;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen break;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen }

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen }

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen return TRUE;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen}

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainenint message_snippet_generate(struct istream *input,

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen unsigned int max_snippet_chars,

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen string_t *snippet)

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen{

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen struct message_parser_ctx *parser;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen struct message_part *parts;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen struct message_decoder_context *decoder;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen struct message_block raw_block, block;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen struct snippet_context ctx;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen pool_t pool;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen int ret;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

efe78d3ba24fc866af1c79b9223dc0809ba26cadStephan Bosch i_zero(&ctx);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen pool = pool_alloconly_create("message snippet", 1024);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx.snippet = snippet;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx.chars_left = max_snippet_chars;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen parser = message_parser_init(pool_datastack_create(), input, 0, 0);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen decoder = message_decoder_init(NULL, 0);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen while ((ret = message_parser_parse_next_block(parser, &raw_block)) > 0) {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (!message_decoder_decode_next_block(decoder, &raw_block, &block))

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen continue;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (block.size == 0) {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen const char *ct;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (block.hdr != NULL)

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen continue;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen /* end of headers - verify that we can use this

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ct = message_decoder_current_content_type(decoder);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (ct == NULL)

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen /* text/plain */ ;

00a50d6dc74e15b5b83afcb8c12c2109a2ca376dTimo Sirainen else if (mail_html2text_content_type_match(ct)) {

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx.html2text = mail_html2text_init(MAIL_HTML2TEXT_FLAG_SKIP_QUOTED);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen ctx.plain_output = buffer_create_dynamic(pool, 1024);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen } else if (strncasecmp(ct, "text/", 5) != 0)

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen break;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen continue;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen }

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen if (!snippet_generate(&ctx, block.data, block.size))

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen break;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen }

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen i_assert(ret != 0);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen message_decoder_deinit(&decoder);

12e5ac049bd74f8b98d9dc62adcb0bf3217beef6Martti Rannanjärvi message_parser_deinit(&parser, &parts);

a2e1929a266a0260fd429fb587588f51d2bd8a96Josef 'Jeff' Sipek mail_html2text_deinit(&ctx.html2text);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen pool_unref(&pool);

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen return input->stream_errno == 0 ? 0 : -1;

7f7be2cbf68f8a202a688d5bc50f82483d461643Timo Sirainen}