simon987/sist2
1
0
Fork 0
mirror of https://github.com/simon987/sist2.git synced 2025-12-12 06:58:54 +00:00
Code Issues Projects Releases Wiki Activity

Add test files as submodule, remove support for msword thumbnails

Browse Source
This commit is contained in:
simon987
2023-02-02 19:52:37 -05:00
parent 87ecc5ef6d
commit 9e0d7bf992
6 changed files with 5 additions and 84 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
third-party/libscan/libscan/msdoc/msdoc.c vendored
View File

@@ -4,8 +4,6 @@
#include <sys/mman.h>
#include "../../third-party/antiword/src/antiword.h"
#include "../ebook/ebook.h"
void parse_msdoc_text(scan_msdoc_ctx_t *ctx, document_t *doc, FILE *file_in, void *buf, size_t buf_len) {
// Open word doc
@@ -71,57 +69,6 @@ void parse_msdoc_text(scan_msdoc_ctx_t *ctx, document_t *doc, FILE *file_in, voi
free(out_buf);
}
void parse_msdoc_pdf(scan_msdoc_ctx_t *ctx, document_t *doc, FILE *file, void *buf, size_t buf_len) {
scan_ebook_ctx_t ebook_ctx = {
.content_size = ctx->content_size,
.tn_size = ctx->tn_size,
.enable_tn = TRUE,
.log = ctx->log,
.logf = ctx->logf,
.store = ctx->store,
};
// Open word doc
options_type *opts = direct_vGetOptions();
opts->iParagraphBreak = 74;
opts->eConversionType = conversion_pdf;
opts->bHideHiddenText = 1;
opts->bRemoveRemovedText = 1;
opts->bUseLandscape = 0;
opts->eEncoding = encoding_latin_1;
opts->iPageHeight = 842; // A4
opts->iPageWidth = 595;
opts->eImageLevel = level_ps_3;
int doc_word_version = iGuessVersionNumber(file, (int) buf_len);
if (doc_word_version < 0 || doc_word_version == 3) {
free(buf);
return;
}
rewind(file);
size_t out_len;
char *out_buf;
FILE *file_out = open_memstream(&out_buf, &out_len);
diagram_type *diag = pCreateDiagram("antiword", NULL, file_out);
if (diag == NULL) {
return;
}
bWordDecryptor(file, (int) buf_len, diag);
vDestroyDiagram(diag);
fclose(file_out);
parse_ebook_mem(&ebook_ctx, out_buf, out_len, "application/pdf", doc, TRUE);
free(buf);
free(out_buf);
}
void parse_msdoc(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc) {
size_t buf_len;
@@ -138,11 +85,6 @@ void parse_msdoc(scan_msdoc_ctx_t *ctx, vfile_t *f, document_t *doc) {
return;
}
if (ctx->enable_tn) {
char *buf_pdf = malloc(buf_len);
memcpy(buf_pdf, buf, buf_len);
parse_msdoc_pdf(ctx, doc, file, buf_pdf, buf_len);
}
parse_msdoc_text(ctx, doc, file, buf, buf_len);
fclose(file);
}

2
third-party/libscan/libscan/msdoc/msdoc.h vendored
View File

@@ -5,8 +5,6 @@
typedef struct {
long content_size;
int enable_tn;
int tn_size;
log_callback_t log;
logf_callback_t logf;
store_callback_t store;