57 lines
1.7 KiB
Markdown
57 lines
1.7 KiB
Markdown
|
|
```c title="C"
|
||
|
|
#include "kreuzberg.h"
|
||
|
|
#include <stdio.h>
|
||
|
|
#include <string.h>
|
||
|
|
|
||
|
|
int main(void) {
|
||
|
|
const char *config_json =
|
||
|
|
"{"
|
||
|
|
"\"language_detection\": {"
|
||
|
|
"\"enabled\": true,"
|
||
|
|
"\"min_confidence\": 0.9,"
|
||
|
|
"\"detect_multiple\": false"
|
||
|
|
"}"
|
||
|
|
"}";
|
||
|
|
|
||
|
|
KREUZBERGExtractionConfig *config = kreuzberg_extraction_config_from_json(config_json);
|
||
|
|
if (!config) {
|
||
|
|
fprintf(stderr, "config parse failed (code %d): %s\n",
|
||
|
|
kreuzberg_last_error_code(),
|
||
|
|
kreuzberg_last_error_context());
|
||
|
|
return 1;
|
||
|
|
}
|
||
|
|
|
||
|
|
KREUZBERGExtractionResult *result =
|
||
|
|
kreuzberg_extract_file_sync("document.pdf", NULL, config);
|
||
|
|
if (!result) {
|
||
|
|
fprintf(stderr, "extraction failed (code %d): %s\n",
|
||
|
|
kreuzberg_last_error_code(),
|
||
|
|
kreuzberg_last_error_context());
|
||
|
|
kreuzberg_extraction_config_free(config);
|
||
|
|
return 1;
|
||
|
|
}
|
||
|
|
|
||
|
|
KREUZBERGMetadata *metadata = kreuzberg_extraction_result_metadata(result);
|
||
|
|
if (metadata) {
|
||
|
|
char *language = kreuzberg_metadata_language(metadata);
|
||
|
|
if (language) {
|
||
|
|
printf("Metadata language: %s\n", language);
|
||
|
|
kreuzberg_free_string(language);
|
||
|
|
}
|
||
|
|
kreuzberg_metadata_free(metadata);
|
||
|
|
}
|
||
|
|
|
||
|
|
char *detected_languages_json = kreuzberg_extraction_result_detected_languages(result);
|
||
|
|
if (detected_languages_json) {
|
||
|
|
printf("Detected languages: %s\n", detected_languages_json);
|
||
|
|
kreuzberg_free_string(detected_languages_json);
|
||
|
|
} else {
|
||
|
|
printf("No languages detected\n");
|
||
|
|
}
|
||
|
|
|
||
|
|
kreuzberg_extraction_result_free(result);
|
||
|
|
kreuzberg_extraction_config_free(config);
|
||
|
|
return 0;
|
||
|
|
}
|
||
|
|
```
|