kreuzberg 4.0.0.rc1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +7 -0
- data/.gitignore +8 -0
- data/.rspec +3 -0
- data/.rubocop.yaml +534 -0
- data/Gemfile +9 -0
- data/Gemfile.lock +157 -0
- data/README.md +421 -0
- data/Rakefile +25 -0
- data/Steepfile +47 -0
- data/examples/async_patterns.rb +340 -0
- data/ext/kreuzberg_rb/extconf.rb +35 -0
- data/ext/kreuzberg_rb/native/Cargo.toml +36 -0
- data/ext/kreuzberg_rb/native/README.md +425 -0
- data/ext/kreuzberg_rb/native/build.rs +17 -0
- data/ext/kreuzberg_rb/native/include/ieeefp.h +11 -0
- data/ext/kreuzberg_rb/native/include/msvc_compat/strings.h +14 -0
- data/ext/kreuzberg_rb/native/include/strings.h +20 -0
- data/ext/kreuzberg_rb/native/include/unistd.h +47 -0
- data/ext/kreuzberg_rb/native/src/lib.rs +2939 -0
- data/extconf.rb +28 -0
- data/kreuzberg.gemspec +105 -0
- data/lib/kreuzberg/api_proxy.rb +142 -0
- data/lib/kreuzberg/cache_api.rb +45 -0
- data/lib/kreuzberg/cli.rb +55 -0
- data/lib/kreuzberg/cli_proxy.rb +127 -0
- data/lib/kreuzberg/config.rb +684 -0
- data/lib/kreuzberg/errors.rb +50 -0
- data/lib/kreuzberg/extraction_api.rb +84 -0
- data/lib/kreuzberg/mcp_proxy.rb +186 -0
- data/lib/kreuzberg/ocr_backend_protocol.rb +113 -0
- data/lib/kreuzberg/post_processor_protocol.rb +86 -0
- data/lib/kreuzberg/result.rb +216 -0
- data/lib/kreuzberg/setup_lib_path.rb +79 -0
- data/lib/kreuzberg/validator_protocol.rb +89 -0
- data/lib/kreuzberg/version.rb +5 -0
- data/lib/kreuzberg.rb +82 -0
- data/pkg/kreuzberg-4.0.0.rc1.gem +0 -0
- data/sig/kreuzberg/internal.rbs +184 -0
- data/sig/kreuzberg.rbs +468 -0
- data/spec/binding/cache_spec.rb +227 -0
- data/spec/binding/cli_proxy_spec.rb +87 -0
- data/spec/binding/cli_spec.rb +54 -0
- data/spec/binding/config_spec.rb +345 -0
- data/spec/binding/config_validation_spec.rb +283 -0
- data/spec/binding/error_handling_spec.rb +213 -0
- data/spec/binding/errors_spec.rb +66 -0
- data/spec/binding/plugins/ocr_backend_spec.rb +307 -0
- data/spec/binding/plugins/postprocessor_spec.rb +269 -0
- data/spec/binding/plugins/validator_spec.rb +274 -0
- data/spec/examples.txt +104 -0
- data/spec/fixtures/config.toml +39 -0
- data/spec/fixtures/config.yaml +42 -0
- data/spec/fixtures/invalid_config.toml +4 -0
- data/spec/smoke/package_spec.rb +178 -0
- data/spec/spec_helper.rb +42 -0
- data/vendor/kreuzberg/Cargo.toml +134 -0
- data/vendor/kreuzberg/README.md +175 -0
- data/vendor/kreuzberg/build.rs +460 -0
- data/vendor/kreuzberg/src/api/error.rs +81 -0
- data/vendor/kreuzberg/src/api/handlers.rs +199 -0
- data/vendor/kreuzberg/src/api/mod.rs +79 -0
- data/vendor/kreuzberg/src/api/server.rs +353 -0
- data/vendor/kreuzberg/src/api/types.rs +170 -0
- data/vendor/kreuzberg/src/bin/profile_extract.rs +455 -0
- data/vendor/kreuzberg/src/cache/mod.rs +1143 -0
- data/vendor/kreuzberg/src/chunking/mod.rs +677 -0
- data/vendor/kreuzberg/src/core/batch_mode.rs +35 -0
- data/vendor/kreuzberg/src/core/config.rs +1032 -0
- data/vendor/kreuzberg/src/core/extractor.rs +903 -0
- data/vendor/kreuzberg/src/core/io.rs +327 -0
- data/vendor/kreuzberg/src/core/mime.rs +615 -0
- data/vendor/kreuzberg/src/core/mod.rs +42 -0
- data/vendor/kreuzberg/src/core/pipeline.rs +906 -0
- data/vendor/kreuzberg/src/embeddings.rs +323 -0
- data/vendor/kreuzberg/src/error.rs +431 -0
- data/vendor/kreuzberg/src/extraction/archive.rs +954 -0
- data/vendor/kreuzberg/src/extraction/docx.rs +40 -0
- data/vendor/kreuzberg/src/extraction/email.rs +854 -0
- data/vendor/kreuzberg/src/extraction/excel.rs +688 -0
- data/vendor/kreuzberg/src/extraction/html.rs +553 -0
- data/vendor/kreuzberg/src/extraction/image.rs +368 -0
- data/vendor/kreuzberg/src/extraction/libreoffice.rs +564 -0
- data/vendor/kreuzberg/src/extraction/mod.rs +77 -0
- data/vendor/kreuzberg/src/extraction/office_metadata/app_properties.rs +398 -0
- data/vendor/kreuzberg/src/extraction/office_metadata/core_properties.rs +247 -0
- data/vendor/kreuzberg/src/extraction/office_metadata/custom_properties.rs +240 -0
- data/vendor/kreuzberg/src/extraction/office_metadata/mod.rs +128 -0
- data/vendor/kreuzberg/src/extraction/pandoc/batch.rs +275 -0
- data/vendor/kreuzberg/src/extraction/pandoc/mime_types.rs +178 -0
- data/vendor/kreuzberg/src/extraction/pandoc/mod.rs +491 -0
- data/vendor/kreuzberg/src/extraction/pandoc/server.rs +496 -0
- data/vendor/kreuzberg/src/extraction/pandoc/subprocess.rs +1188 -0
- data/vendor/kreuzberg/src/extraction/pandoc/version.rs +162 -0
- data/vendor/kreuzberg/src/extraction/pptx.rs +3000 -0
- data/vendor/kreuzberg/src/extraction/structured.rs +490 -0
- data/vendor/kreuzberg/src/extraction/table.rs +328 -0
- data/vendor/kreuzberg/src/extraction/text.rs +269 -0
- data/vendor/kreuzberg/src/extraction/xml.rs +333 -0
- data/vendor/kreuzberg/src/extractors/archive.rs +425 -0
- data/vendor/kreuzberg/src/extractors/docx.rs +479 -0
- data/vendor/kreuzberg/src/extractors/email.rs +129 -0
- data/vendor/kreuzberg/src/extractors/excel.rs +344 -0
- data/vendor/kreuzberg/src/extractors/html.rs +410 -0
- data/vendor/kreuzberg/src/extractors/image.rs +195 -0
- data/vendor/kreuzberg/src/extractors/mod.rs +268 -0
- data/vendor/kreuzberg/src/extractors/pandoc.rs +201 -0
- data/vendor/kreuzberg/src/extractors/pdf.rs +496 -0
- data/vendor/kreuzberg/src/extractors/pptx.rs +234 -0
- data/vendor/kreuzberg/src/extractors/structured.rs +126 -0
- data/vendor/kreuzberg/src/extractors/text.rs +242 -0
- data/vendor/kreuzberg/src/extractors/xml.rs +128 -0
- data/vendor/kreuzberg/src/image/dpi.rs +164 -0
- data/vendor/kreuzberg/src/image/mod.rs +6 -0
- data/vendor/kreuzberg/src/image/preprocessing.rs +417 -0
- data/vendor/kreuzberg/src/image/resize.rs +89 -0
- data/vendor/kreuzberg/src/keywords/config.rs +154 -0
- data/vendor/kreuzberg/src/keywords/mod.rs +237 -0
- data/vendor/kreuzberg/src/keywords/processor.rs +267 -0
- data/vendor/kreuzberg/src/keywords/rake.rs +294 -0
- data/vendor/kreuzberg/src/keywords/types.rs +68 -0
- data/vendor/kreuzberg/src/keywords/yake.rs +163 -0
- data/vendor/kreuzberg/src/language_detection/mod.rs +942 -0
- data/vendor/kreuzberg/src/lib.rs +102 -0
- data/vendor/kreuzberg/src/mcp/mod.rs +32 -0
- data/vendor/kreuzberg/src/mcp/server.rs +1966 -0
- data/vendor/kreuzberg/src/ocr/cache.rs +469 -0
- data/vendor/kreuzberg/src/ocr/error.rs +37 -0
- data/vendor/kreuzberg/src/ocr/hocr.rs +216 -0
- data/vendor/kreuzberg/src/ocr/mod.rs +58 -0
- data/vendor/kreuzberg/src/ocr/processor.rs +847 -0
- data/vendor/kreuzberg/src/ocr/table/mod.rs +4 -0
- data/vendor/kreuzberg/src/ocr/table/tsv_parser.rs +144 -0
- data/vendor/kreuzberg/src/ocr/tesseract_backend.rs +450 -0
- data/vendor/kreuzberg/src/ocr/types.rs +393 -0
- data/vendor/kreuzberg/src/ocr/utils.rs +47 -0
- data/vendor/kreuzberg/src/ocr/validation.rs +206 -0
- data/vendor/kreuzberg/src/pdf/error.rs +122 -0
- data/vendor/kreuzberg/src/pdf/images.rs +139 -0
- data/vendor/kreuzberg/src/pdf/metadata.rs +346 -0
- data/vendor/kreuzberg/src/pdf/mod.rs +50 -0
- data/vendor/kreuzberg/src/pdf/rendering.rs +369 -0
- data/vendor/kreuzberg/src/pdf/table.rs +420 -0
- data/vendor/kreuzberg/src/pdf/text.rs +161 -0
- data/vendor/kreuzberg/src/plugins/extractor.rs +1010 -0
- data/vendor/kreuzberg/src/plugins/mod.rs +209 -0
- data/vendor/kreuzberg/src/plugins/ocr.rs +629 -0
- data/vendor/kreuzberg/src/plugins/processor.rs +641 -0
- data/vendor/kreuzberg/src/plugins/registry.rs +1324 -0
- data/vendor/kreuzberg/src/plugins/traits.rs +258 -0
- data/vendor/kreuzberg/src/plugins/validator.rs +955 -0
- data/vendor/kreuzberg/src/stopwords/mod.rs +1470 -0
- data/vendor/kreuzberg/src/text/mod.rs +19 -0
- data/vendor/kreuzberg/src/text/quality.rs +697 -0
- data/vendor/kreuzberg/src/text/string_utils.rs +217 -0
- data/vendor/kreuzberg/src/text/token_reduction/cjk_utils.rs +164 -0
- data/vendor/kreuzberg/src/text/token_reduction/config.rs +100 -0
- data/vendor/kreuzberg/src/text/token_reduction/core.rs +796 -0
- data/vendor/kreuzberg/src/text/token_reduction/filters.rs +902 -0
- data/vendor/kreuzberg/src/text/token_reduction/mod.rs +160 -0
- data/vendor/kreuzberg/src/text/token_reduction/semantic.rs +619 -0
- data/vendor/kreuzberg/src/text/token_reduction/simd_text.rs +147 -0
- data/vendor/kreuzberg/src/types.rs +873 -0
- data/vendor/kreuzberg/src/utils/mod.rs +17 -0
- data/vendor/kreuzberg/src/utils/quality.rs +959 -0
- data/vendor/kreuzberg/src/utils/string_utils.rs +381 -0
- data/vendor/kreuzberg/stopwords/af_stopwords.json +53 -0
- data/vendor/kreuzberg/stopwords/ar_stopwords.json +482 -0
- data/vendor/kreuzberg/stopwords/bg_stopwords.json +261 -0
- data/vendor/kreuzberg/stopwords/bn_stopwords.json +400 -0
- data/vendor/kreuzberg/stopwords/br_stopwords.json +1205 -0
- data/vendor/kreuzberg/stopwords/ca_stopwords.json +280 -0
- data/vendor/kreuzberg/stopwords/cs_stopwords.json +425 -0
- data/vendor/kreuzberg/stopwords/da_stopwords.json +172 -0
- data/vendor/kreuzberg/stopwords/de_stopwords.json +622 -0
- data/vendor/kreuzberg/stopwords/el_stopwords.json +849 -0
- data/vendor/kreuzberg/stopwords/en_stopwords.json +1300 -0
- data/vendor/kreuzberg/stopwords/eo_stopwords.json +175 -0
- data/vendor/kreuzberg/stopwords/es_stopwords.json +734 -0
- data/vendor/kreuzberg/stopwords/et_stopwords.json +37 -0
- data/vendor/kreuzberg/stopwords/eu_stopwords.json +100 -0
- data/vendor/kreuzberg/stopwords/fa_stopwords.json +801 -0
- data/vendor/kreuzberg/stopwords/fi_stopwords.json +849 -0
- data/vendor/kreuzberg/stopwords/fr_stopwords.json +693 -0
- data/vendor/kreuzberg/stopwords/ga_stopwords.json +111 -0
- data/vendor/kreuzberg/stopwords/gl_stopwords.json +162 -0
- data/vendor/kreuzberg/stopwords/gu_stopwords.json +226 -0
- data/vendor/kreuzberg/stopwords/ha_stopwords.json +41 -0
- data/vendor/kreuzberg/stopwords/he_stopwords.json +196 -0
- data/vendor/kreuzberg/stopwords/hi_stopwords.json +227 -0
- data/vendor/kreuzberg/stopwords/hr_stopwords.json +181 -0
- data/vendor/kreuzberg/stopwords/hu_stopwords.json +791 -0
- data/vendor/kreuzberg/stopwords/hy_stopwords.json +47 -0
- data/vendor/kreuzberg/stopwords/id_stopwords.json +760 -0
- data/vendor/kreuzberg/stopwords/it_stopwords.json +634 -0
- data/vendor/kreuzberg/stopwords/ja_stopwords.json +136 -0
- data/vendor/kreuzberg/stopwords/kn_stopwords.json +84 -0
- data/vendor/kreuzberg/stopwords/ko_stopwords.json +681 -0
- data/vendor/kreuzberg/stopwords/ku_stopwords.json +64 -0
- data/vendor/kreuzberg/stopwords/la_stopwords.json +51 -0
- data/vendor/kreuzberg/stopwords/lt_stopwords.json +476 -0
- data/vendor/kreuzberg/stopwords/lv_stopwords.json +163 -0
- data/vendor/kreuzberg/stopwords/ml_stopwords.json +1 -0
- data/vendor/kreuzberg/stopwords/mr_stopwords.json +101 -0
- data/vendor/kreuzberg/stopwords/ms_stopwords.json +477 -0
- data/vendor/kreuzberg/stopwords/ne_stopwords.json +490 -0
- data/vendor/kreuzberg/stopwords/nl_stopwords.json +415 -0
- data/vendor/kreuzberg/stopwords/no_stopwords.json +223 -0
- data/vendor/kreuzberg/stopwords/pl_stopwords.json +331 -0
- data/vendor/kreuzberg/stopwords/pt_stopwords.json +562 -0
- data/vendor/kreuzberg/stopwords/ro_stopwords.json +436 -0
- data/vendor/kreuzberg/stopwords/ru_stopwords.json +561 -0
- data/vendor/kreuzberg/stopwords/si_stopwords.json +193 -0
- data/vendor/kreuzberg/stopwords/sk_stopwords.json +420 -0
- data/vendor/kreuzberg/stopwords/sl_stopwords.json +448 -0
- data/vendor/kreuzberg/stopwords/so_stopwords.json +32 -0
- data/vendor/kreuzberg/stopwords/st_stopwords.json +33 -0
- data/vendor/kreuzberg/stopwords/sv_stopwords.json +420 -0
- data/vendor/kreuzberg/stopwords/sw_stopwords.json +76 -0
- data/vendor/kreuzberg/stopwords/ta_stopwords.json +129 -0
- data/vendor/kreuzberg/stopwords/te_stopwords.json +54 -0
- data/vendor/kreuzberg/stopwords/th_stopwords.json +118 -0
- data/vendor/kreuzberg/stopwords/tl_stopwords.json +149 -0
- data/vendor/kreuzberg/stopwords/tr_stopwords.json +506 -0
- data/vendor/kreuzberg/stopwords/uk_stopwords.json +75 -0
- data/vendor/kreuzberg/stopwords/ur_stopwords.json +519 -0
- data/vendor/kreuzberg/stopwords/vi_stopwords.json +647 -0
- data/vendor/kreuzberg/stopwords/yo_stopwords.json +62 -0
- data/vendor/kreuzberg/stopwords/zh_stopwords.json +796 -0
- data/vendor/kreuzberg/stopwords/zu_stopwords.json +31 -0
- data/vendor/kreuzberg/tests/api_tests.rs +966 -0
- data/vendor/kreuzberg/tests/archive_integration.rs +543 -0
- data/vendor/kreuzberg/tests/batch_orchestration.rs +542 -0
- data/vendor/kreuzberg/tests/batch_processing.rs +304 -0
- data/vendor/kreuzberg/tests/chunking_offset_demo.rs +92 -0
- data/vendor/kreuzberg/tests/concurrency_stress.rs +509 -0
- data/vendor/kreuzberg/tests/config_features.rs +580 -0
- data/vendor/kreuzberg/tests/config_loading_tests.rs +439 -0
- data/vendor/kreuzberg/tests/core_integration.rs +493 -0
- data/vendor/kreuzberg/tests/csv_integration.rs +424 -0
- data/vendor/kreuzberg/tests/docx_metadata_extraction_test.rs +124 -0
- data/vendor/kreuzberg/tests/email_integration.rs +325 -0
- data/vendor/kreuzberg/tests/error_handling.rs +393 -0
- data/vendor/kreuzberg/tests/format_integration.rs +159 -0
- data/vendor/kreuzberg/tests/helpers/mod.rs +142 -0
- data/vendor/kreuzberg/tests/image_integration.rs +253 -0
- data/vendor/kreuzberg/tests/keywords_integration.rs +479 -0
- data/vendor/kreuzberg/tests/keywords_quality.rs +509 -0
- data/vendor/kreuzberg/tests/mime_detection.rs +428 -0
- data/vendor/kreuzberg/tests/ocr_configuration.rs +510 -0
- data/vendor/kreuzberg/tests/ocr_errors.rs +676 -0
- data/vendor/kreuzberg/tests/ocr_quality.rs +627 -0
- data/vendor/kreuzberg/tests/ocr_stress.rs +469 -0
- data/vendor/kreuzberg/tests/pandoc_integration.rs +503 -0
- data/vendor/kreuzberg/tests/pdf_integration.rs +43 -0
- data/vendor/kreuzberg/tests/pipeline_integration.rs +1412 -0
- data/vendor/kreuzberg/tests/plugin_ocr_backend_test.rs +771 -0
- data/vendor/kreuzberg/tests/plugin_postprocessor_test.rs +561 -0
- data/vendor/kreuzberg/tests/plugin_system.rs +921 -0
- data/vendor/kreuzberg/tests/plugin_validator_test.rs +783 -0
- data/vendor/kreuzberg/tests/registry_integration_tests.rs +607 -0
- data/vendor/kreuzberg/tests/security_validation.rs +404 -0
- data/vendor/kreuzberg/tests/stopwords_integration_test.rs +888 -0
- data/vendor/kreuzberg/tests/test_fastembed.rs +609 -0
- data/vendor/kreuzberg/tests/xlsx_metadata_extraction_test.rs +87 -0
- metadata +471 -0
|
@@ -0,0 +1,368 @@
|
|
|
1
|
+
//! Image extraction functionality.
|
|
2
|
+
//!
|
|
3
|
+
//! This module provides functions for extracting metadata and EXIF data from images.
|
|
4
|
+
|
|
5
|
+
use crate::error::{KreuzbergError, Result};
|
|
6
|
+
use exif::{In, Reader, Tag};
|
|
7
|
+
use image::ImageReader;
|
|
8
|
+
use std::collections::HashMap;
|
|
9
|
+
use std::io::Cursor;
|
|
10
|
+
|
|
11
|
+
/// Image metadata extracted from an image file.
|
|
12
|
+
#[derive(Debug, Clone)]
|
|
13
|
+
pub struct ImageMetadata {
|
|
14
|
+
/// Image width in pixels
|
|
15
|
+
pub width: u32,
|
|
16
|
+
/// Image height in pixels
|
|
17
|
+
pub height: u32,
|
|
18
|
+
/// Image format (e.g., "PNG", "JPEG")
|
|
19
|
+
pub format: String,
|
|
20
|
+
/// EXIF data if available
|
|
21
|
+
pub exif_data: HashMap<String, String>,
|
|
22
|
+
}
|
|
23
|
+
|
|
24
|
+
/// Extract metadata from image bytes.
|
|
25
|
+
///
|
|
26
|
+
/// Extracts dimensions, format, and EXIF data from the image.
|
|
27
|
+
pub fn extract_image_metadata(bytes: &[u8]) -> Result<ImageMetadata> {
|
|
28
|
+
let reader = ImageReader::new(Cursor::new(bytes))
|
|
29
|
+
.with_guessed_format()
|
|
30
|
+
.map_err(|e| KreuzbergError::parsing(format!("Failed to read image format: {}", e)))?;
|
|
31
|
+
|
|
32
|
+
let format = reader
|
|
33
|
+
.format()
|
|
34
|
+
.ok_or_else(|| KreuzbergError::parsing("Could not determine image format".to_string()))?;
|
|
35
|
+
|
|
36
|
+
let image = reader
|
|
37
|
+
.decode()
|
|
38
|
+
.map_err(|e| KreuzbergError::parsing(format!("Failed to decode image: {}", e)))?;
|
|
39
|
+
|
|
40
|
+
let width = image.width();
|
|
41
|
+
let height = image.height();
|
|
42
|
+
let format_str = format!("{:?}", format);
|
|
43
|
+
|
|
44
|
+
let exif_data = extract_exif_data(bytes);
|
|
45
|
+
|
|
46
|
+
Ok(ImageMetadata {
|
|
47
|
+
width,
|
|
48
|
+
height,
|
|
49
|
+
format: format_str,
|
|
50
|
+
exif_data,
|
|
51
|
+
})
|
|
52
|
+
}
|
|
53
|
+
|
|
54
|
+
/// Extract EXIF data from image bytes.
|
|
55
|
+
///
|
|
56
|
+
/// Returns a HashMap of EXIF tags and their values.
|
|
57
|
+
/// If EXIF data is not available or cannot be parsed, returns an empty HashMap.
|
|
58
|
+
fn extract_exif_data(bytes: &[u8]) -> HashMap<String, String> {
|
|
59
|
+
let mut exif_map = HashMap::new();
|
|
60
|
+
|
|
61
|
+
let exif_reader = match Reader::new().read_from_container(&mut Cursor::new(bytes)) {
|
|
62
|
+
Ok(reader) => reader,
|
|
63
|
+
Err(_) => return exif_map,
|
|
64
|
+
};
|
|
65
|
+
|
|
66
|
+
let common_tags = [
|
|
67
|
+
(Tag::Make, "Make"),
|
|
68
|
+
(Tag::Model, "Model"),
|
|
69
|
+
(Tag::DateTime, "DateTime"),
|
|
70
|
+
(Tag::DateTimeOriginal, "DateTimeOriginal"),
|
|
71
|
+
(Tag::DateTimeDigitized, "DateTimeDigitized"),
|
|
72
|
+
(Tag::Software, "Software"),
|
|
73
|
+
(Tag::Orientation, "Orientation"),
|
|
74
|
+
(Tag::XResolution, "XResolution"),
|
|
75
|
+
(Tag::YResolution, "YResolution"),
|
|
76
|
+
(Tag::ResolutionUnit, "ResolutionUnit"),
|
|
77
|
+
(Tag::ExposureTime, "ExposureTime"),
|
|
78
|
+
(Tag::FNumber, "FNumber"),
|
|
79
|
+
(Tag::PhotographicSensitivity, "ISO"),
|
|
80
|
+
(Tag::FocalLength, "FocalLength"),
|
|
81
|
+
(Tag::Flash, "Flash"),
|
|
82
|
+
(Tag::WhiteBalance, "WhiteBalance"),
|
|
83
|
+
(Tag::GPSLatitude, "GPSLatitude"),
|
|
84
|
+
(Tag::GPSLongitude, "GPSLongitude"),
|
|
85
|
+
(Tag::GPSAltitude, "GPSAltitude"),
|
|
86
|
+
];
|
|
87
|
+
|
|
88
|
+
for (tag, field_name) in common_tags {
|
|
89
|
+
if let Some(field) = exif_reader.get_field(tag, In::PRIMARY) {
|
|
90
|
+
exif_map.insert(field_name.to_string(), field.display_value().to_string());
|
|
91
|
+
}
|
|
92
|
+
}
|
|
93
|
+
|
|
94
|
+
exif_map
|
|
95
|
+
}
|
|
96
|
+
|
|
97
|
+
#[cfg(test)]
|
|
98
|
+
mod tests {
|
|
99
|
+
use super::*;
|
|
100
|
+
use image::{ImageBuffer, ImageFormat, Rgb, RgbImage};
|
|
101
|
+
use std::io::Cursor;
|
|
102
|
+
|
|
103
|
+
fn create_test_image(width: u32, height: u32, format: ImageFormat) -> Vec<u8> {
|
|
104
|
+
let img: RgbImage = ImageBuffer::from_fn(width, height, |x, y| {
|
|
105
|
+
let r = ((x as f32 / width as f32) * 255.0) as u8;
|
|
106
|
+
let g = ((y as f32 / height as f32) * 255.0) as u8;
|
|
107
|
+
let b = 128;
|
|
108
|
+
Rgb([r, g, b])
|
|
109
|
+
});
|
|
110
|
+
|
|
111
|
+
let mut bytes: Vec<u8> = Vec::new();
|
|
112
|
+
let mut cursor = Cursor::new(&mut bytes);
|
|
113
|
+
img.write_to(&mut cursor, format).unwrap();
|
|
114
|
+
bytes
|
|
115
|
+
}
|
|
116
|
+
|
|
117
|
+
#[test]
|
|
118
|
+
fn test_extract_png_image_returns_correct_metadata() {
|
|
119
|
+
let bytes = create_test_image(100, 80, ImageFormat::Png);
|
|
120
|
+
let result = extract_image_metadata(&bytes);
|
|
121
|
+
|
|
122
|
+
assert!(result.is_ok());
|
|
123
|
+
let metadata = result.unwrap();
|
|
124
|
+
assert_eq!(metadata.width, 100);
|
|
125
|
+
assert_eq!(metadata.height, 80);
|
|
126
|
+
assert_eq!(metadata.format, "Png");
|
|
127
|
+
}
|
|
128
|
+
|
|
129
|
+
#[test]
|
|
130
|
+
fn test_extract_jpeg_image_returns_correct_metadata() {
|
|
131
|
+
let bytes = create_test_image(200, 150, ImageFormat::Jpeg);
|
|
132
|
+
let result = extract_image_metadata(&bytes);
|
|
133
|
+
|
|
134
|
+
assert!(result.is_ok());
|
|
135
|
+
let metadata = result.unwrap();
|
|
136
|
+
assert_eq!(metadata.width, 200);
|
|
137
|
+
assert_eq!(metadata.height, 150);
|
|
138
|
+
assert_eq!(metadata.format, "Jpeg");
|
|
139
|
+
}
|
|
140
|
+
|
|
141
|
+
#[test]
|
|
142
|
+
fn test_extract_webp_image_returns_correct_metadata() {
|
|
143
|
+
let bytes = create_test_image(120, 90, ImageFormat::WebP);
|
|
144
|
+
let result = extract_image_metadata(&bytes);
|
|
145
|
+
|
|
146
|
+
assert!(result.is_ok());
|
|
147
|
+
let metadata = result.unwrap();
|
|
148
|
+
assert_eq!(metadata.width, 120);
|
|
149
|
+
assert_eq!(metadata.height, 90);
|
|
150
|
+
assert_eq!(metadata.format, "WebP");
|
|
151
|
+
}
|
|
152
|
+
|
|
153
|
+
#[test]
|
|
154
|
+
fn test_extract_bmp_image_returns_correct_metadata() {
|
|
155
|
+
let bytes = create_test_image(50, 50, ImageFormat::Bmp);
|
|
156
|
+
let result = extract_image_metadata(&bytes);
|
|
157
|
+
|
|
158
|
+
assert!(result.is_ok());
|
|
159
|
+
let metadata = result.unwrap();
|
|
160
|
+
assert_eq!(metadata.width, 50);
|
|
161
|
+
assert_eq!(metadata.height, 50);
|
|
162
|
+
assert_eq!(metadata.format, "Bmp");
|
|
163
|
+
}
|
|
164
|
+
|
|
165
|
+
#[test]
|
|
166
|
+
fn test_extract_tiff_image_returns_correct_metadata() {
|
|
167
|
+
let bytes = create_test_image(180, 120, ImageFormat::Tiff);
|
|
168
|
+
let result = extract_image_metadata(&bytes);
|
|
169
|
+
|
|
170
|
+
assert!(result.is_ok());
|
|
171
|
+
let metadata = result.unwrap();
|
|
172
|
+
assert_eq!(metadata.width, 180);
|
|
173
|
+
assert_eq!(metadata.height, 120);
|
|
174
|
+
assert_eq!(metadata.format, "Tiff");
|
|
175
|
+
}
|
|
176
|
+
|
|
177
|
+
#[test]
|
|
178
|
+
fn test_extract_gif_image_returns_correct_metadata() {
|
|
179
|
+
let bytes = create_test_image(64, 64, ImageFormat::Gif);
|
|
180
|
+
let result = extract_image_metadata(&bytes);
|
|
181
|
+
|
|
182
|
+
assert!(result.is_ok());
|
|
183
|
+
let metadata = result.unwrap();
|
|
184
|
+
assert_eq!(metadata.width, 64);
|
|
185
|
+
assert_eq!(metadata.height, 64);
|
|
186
|
+
assert_eq!(metadata.format, "Gif");
|
|
187
|
+
}
|
|
188
|
+
|
|
189
|
+
#[test]
|
|
190
|
+
fn test_extract_image_extreme_aspect_ratio() {
|
|
191
|
+
let bytes = create_test_image(1000, 10, ImageFormat::Png);
|
|
192
|
+
let result = extract_image_metadata(&bytes);
|
|
193
|
+
|
|
194
|
+
assert!(result.is_ok());
|
|
195
|
+
let metadata = result.unwrap();
|
|
196
|
+
assert_eq!(metadata.width, 1000);
|
|
197
|
+
assert_eq!(metadata.height, 10);
|
|
198
|
+
assert!(metadata.width / metadata.height >= 100);
|
|
199
|
+
}
|
|
200
|
+
|
|
201
|
+
#[test]
|
|
202
|
+
fn test_extract_image_dimensions_correctly() {
|
|
203
|
+
let bytes = create_test_image(640, 480, ImageFormat::Png);
|
|
204
|
+
let result = extract_image_metadata(&bytes);
|
|
205
|
+
|
|
206
|
+
assert!(result.is_ok());
|
|
207
|
+
let metadata = result.unwrap();
|
|
208
|
+
assert_eq!(metadata.width, 640);
|
|
209
|
+
assert_eq!(metadata.height, 480);
|
|
210
|
+
}
|
|
211
|
+
|
|
212
|
+
#[test]
|
|
213
|
+
fn test_extract_image_format_correctly() {
|
|
214
|
+
let png_bytes = create_test_image(100, 100, ImageFormat::Png);
|
|
215
|
+
let jpeg_bytes = create_test_image(100, 100, ImageFormat::Jpeg);
|
|
216
|
+
|
|
217
|
+
let png_metadata = extract_image_metadata(&png_bytes).unwrap();
|
|
218
|
+
let jpeg_metadata = extract_image_metadata(&jpeg_bytes).unwrap();
|
|
219
|
+
|
|
220
|
+
assert_eq!(png_metadata.format, "Png");
|
|
221
|
+
assert_eq!(jpeg_metadata.format, "Jpeg");
|
|
222
|
+
}
|
|
223
|
+
|
|
224
|
+
#[test]
|
|
225
|
+
fn test_extract_image_without_exif_returns_empty_map() {
|
|
226
|
+
let bytes = create_test_image(100, 100, ImageFormat::Png);
|
|
227
|
+
let result = extract_image_metadata(&bytes);
|
|
228
|
+
|
|
229
|
+
assert!(result.is_ok());
|
|
230
|
+
let metadata = result.unwrap();
|
|
231
|
+
assert!(metadata.exif_data.is_empty());
|
|
232
|
+
}
|
|
233
|
+
|
|
234
|
+
#[test]
|
|
235
|
+
fn test_extract_exif_data_from_jpeg_with_exif() {
|
|
236
|
+
let bytes = create_test_image(100, 100, ImageFormat::Jpeg);
|
|
237
|
+
let result = extract_image_metadata(&bytes);
|
|
238
|
+
|
|
239
|
+
assert!(result.is_ok());
|
|
240
|
+
let metadata = result.unwrap();
|
|
241
|
+
assert_eq!(metadata.exif_data.len(), 0);
|
|
242
|
+
}
|
|
243
|
+
|
|
244
|
+
#[test]
|
|
245
|
+
fn test_extract_image_metadata_invalid_returns_error() {
|
|
246
|
+
let invalid_bytes = vec![0, 1, 2, 3, 4, 5];
|
|
247
|
+
let result = extract_image_metadata(&invalid_bytes);
|
|
248
|
+
assert!(result.is_err());
|
|
249
|
+
}
|
|
250
|
+
|
|
251
|
+
#[test]
|
|
252
|
+
fn test_extract_image_corrupted_data_returns_error() {
|
|
253
|
+
let mut bytes = create_test_image(100, 100, ImageFormat::Png);
|
|
254
|
+
if bytes.len() > 50 {
|
|
255
|
+
for byte in bytes.iter_mut().take(50).skip(20) {
|
|
256
|
+
*byte = 0xFF;
|
|
257
|
+
}
|
|
258
|
+
}
|
|
259
|
+
|
|
260
|
+
let result = extract_image_metadata(&bytes);
|
|
261
|
+
assert!(result.is_ok() || result.is_err());
|
|
262
|
+
}
|
|
263
|
+
|
|
264
|
+
#[test]
|
|
265
|
+
fn test_extract_image_empty_bytes_returns_error() {
|
|
266
|
+
let empty_bytes: Vec<u8> = Vec::new();
|
|
267
|
+
let result = extract_image_metadata(&empty_bytes);
|
|
268
|
+
assert!(result.is_err());
|
|
269
|
+
}
|
|
270
|
+
|
|
271
|
+
#[test]
|
|
272
|
+
fn test_extract_image_unsupported_format_returns_error() {
|
|
273
|
+
let unsupported_bytes = vec![0x00, 0x00, 0x00, 0x0C, 0x6A, 0x50, 0x20, 0x20, 0x0D, 0x0A, 0x87, 0x0A];
|
|
274
|
+
let result = extract_image_metadata(&unsupported_bytes);
|
|
275
|
+
assert!(result.is_err());
|
|
276
|
+
}
|
|
277
|
+
|
|
278
|
+
#[test]
|
|
279
|
+
fn test_extract_very_small_image_1x1_pixel() {
|
|
280
|
+
let bytes = create_test_image(1, 1, ImageFormat::Png);
|
|
281
|
+
let result = extract_image_metadata(&bytes);
|
|
282
|
+
|
|
283
|
+
assert!(result.is_ok());
|
|
284
|
+
let metadata = result.unwrap();
|
|
285
|
+
assert_eq!(metadata.width, 1);
|
|
286
|
+
assert_eq!(metadata.height, 1);
|
|
287
|
+
assert_eq!(metadata.format, "Png");
|
|
288
|
+
}
|
|
289
|
+
|
|
290
|
+
#[test]
|
|
291
|
+
fn test_extract_large_image_dimensions() {
|
|
292
|
+
let bytes = create_test_image(2048, 1536, ImageFormat::Png);
|
|
293
|
+
let result = extract_image_metadata(&bytes);
|
|
294
|
+
|
|
295
|
+
assert!(result.is_ok());
|
|
296
|
+
let metadata = result.unwrap();
|
|
297
|
+
assert_eq!(metadata.width, 2048);
|
|
298
|
+
assert_eq!(metadata.height, 1536);
|
|
299
|
+
}
|
|
300
|
+
|
|
301
|
+
#[test]
|
|
302
|
+
fn test_extract_image_with_no_metadata_has_empty_exif() {
|
|
303
|
+
let bytes = create_test_image(100, 100, ImageFormat::Png);
|
|
304
|
+
let result = extract_image_metadata(&bytes);
|
|
305
|
+
|
|
306
|
+
assert!(result.is_ok());
|
|
307
|
+
let metadata = result.unwrap();
|
|
308
|
+
assert!(metadata.exif_data.is_empty());
|
|
309
|
+
}
|
|
310
|
+
|
|
311
|
+
#[test]
|
|
312
|
+
fn test_extract_exif_data_returns_empty_map_for_non_jpeg() {
|
|
313
|
+
let png_bytes = create_test_image(100, 100, ImageFormat::Png);
|
|
314
|
+
let exif_data = extract_exif_data(&png_bytes);
|
|
315
|
+
assert!(exif_data.is_empty());
|
|
316
|
+
}
|
|
317
|
+
|
|
318
|
+
#[test]
|
|
319
|
+
fn test_extract_rectangular_image_portrait_orientation() {
|
|
320
|
+
let bytes = create_test_image(400, 800, ImageFormat::Jpeg);
|
|
321
|
+
let result = extract_image_metadata(&bytes);
|
|
322
|
+
|
|
323
|
+
assert!(result.is_ok());
|
|
324
|
+
let metadata = result.unwrap();
|
|
325
|
+
assert_eq!(metadata.width, 400);
|
|
326
|
+
assert_eq!(metadata.height, 800);
|
|
327
|
+
assert!(metadata.height > metadata.width);
|
|
328
|
+
}
|
|
329
|
+
|
|
330
|
+
#[test]
|
|
331
|
+
fn test_extract_rectangular_image_landscape_orientation() {
|
|
332
|
+
let bytes = create_test_image(800, 400, ImageFormat::Png);
|
|
333
|
+
let result = extract_image_metadata(&bytes);
|
|
334
|
+
|
|
335
|
+
assert!(result.is_ok());
|
|
336
|
+
let metadata = result.unwrap();
|
|
337
|
+
assert_eq!(metadata.width, 800);
|
|
338
|
+
assert_eq!(metadata.height, 400);
|
|
339
|
+
assert!(metadata.width > metadata.height);
|
|
340
|
+
}
|
|
341
|
+
|
|
342
|
+
#[test]
|
|
343
|
+
fn test_extract_square_image_equal_dimensions() {
|
|
344
|
+
let bytes = create_test_image(512, 512, ImageFormat::Png);
|
|
345
|
+
let result = extract_image_metadata(&bytes);
|
|
346
|
+
|
|
347
|
+
assert!(result.is_ok());
|
|
348
|
+
let metadata = result.unwrap();
|
|
349
|
+
assert_eq!(metadata.width, 512);
|
|
350
|
+
assert_eq!(metadata.height, 512);
|
|
351
|
+
assert_eq!(metadata.width, metadata.height);
|
|
352
|
+
}
|
|
353
|
+
|
|
354
|
+
#[test]
|
|
355
|
+
fn test_extract_metadata_preserves_format_case() {
|
|
356
|
+
let png_bytes = create_test_image(100, 100, ImageFormat::Png);
|
|
357
|
+
let jpeg_bytes = create_test_image(100, 100, ImageFormat::Jpeg);
|
|
358
|
+
let webp_bytes = create_test_image(100, 100, ImageFormat::WebP);
|
|
359
|
+
|
|
360
|
+
let png_meta = extract_image_metadata(&png_bytes).unwrap();
|
|
361
|
+
let jpeg_meta = extract_image_metadata(&jpeg_bytes).unwrap();
|
|
362
|
+
let webp_meta = extract_image_metadata(&webp_bytes).unwrap();
|
|
363
|
+
|
|
364
|
+
assert_eq!(png_meta.format, "Png");
|
|
365
|
+
assert_eq!(jpeg_meta.format, "Jpeg");
|
|
366
|
+
assert_eq!(webp_meta.format, "WebP");
|
|
367
|
+
}
|
|
368
|
+
}
|