Merge pull request #3 from occ-ai/roy.add_output_formatting

Add vendor/inja and output formatting support

.gitignore

@@ -15,6 +15,8 @@
 !CMakePresets.json
 !LICENSE
 !README.md
+!/vendor
+!/vendor/inja
 
 # Exclude lock files
 *.lock.json

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "vendor/inja"]
+	path = vendor/inja
+	url = [email protected]:pantor/inja.git

CMakeLists.txt

@@ -53,7 +53,10 @@ endif()
 include(cmake/FetchTesseract.cmake)
 target_link_libraries(${CMAKE_PROJECT_NAME} PRIVATE Tesseract)
 
+include(cmake/BuildInja.cmake)
+target_link_libraries(${CMAKE_PROJECT_NAME} PRIVATE inja)
+
 target_sources(${CMAKE_PROJECT_NAME} PRIVATE src/plugin-main.c src/obs-utils.cpp src/tesseract-ocr-utils.cpp
-src/ocr-filter.cpp src/ocr-filter-info.c)
+                                             src/ocr-filter.cpp src/ocr-filter-info.c)
 
 set_target_properties_plugin(${CMAKE_PROJECT_NAME} PROPERTIES OUTPUT_NAME ${_name})

CMakePresets.json

@@ -92,7 +92,7 @@
       "description": "Build for Linux x86_64 on CI",
       "cacheVariables": {
         "CMAKE_BUILD_TYPE": "RelWithDebInfo",
-        "CMAKE_COMPILE_WARNING_AS_ERROR": true
+        "CMAKE_COMPILE_WARNING_AS_ERROR": false
       }
     },
     {
@@ -121,7 +121,7 @@
       "description": "Build for Linux aarch64 on CI",
       "cacheVariables": {
         "CMAKE_BUILD_TYPE": "RelWithDebInfo",
-        "CMAKE_COMPILE_WARNING_AS_ERROR": true
+        "CMAKE_COMPILE_WARNING_AS_ERROR": false
       }
     }
   ],

README.md

@@ -49,14 +49,16 @@ Available now:
  - Choose the segmentation mode: Word, Line, Page, etc.
  - "Semantic Smoothing": getting more consistent outputs with higher accuracy and confidence by "averaging" several text outputs
  - Timing/Running modes: per X-milliseconds
+ - Output formatting (with inja): e.g. "Score: {{score}}"
 
 Coming soon:
  - More languages built-in (pretrained Tesseract models)
  - Allowing external model files
- - More output capabilities: Parsing, formatting
+ - More output capabilities e.g. Parsing
  - Output to more built-in OBS sources (Image, etc.)
  - Extracting text from complex image layouts
- - Different timing/run modes: per X-frames, source change, etc.
+ - Different timing/run modes: per X-frames, image change, etc.
+ - Image stabilization
 
 Check out our other plugins:
 - [Background Removal](https://github.com/occ-ai/obs-backgroundremoval) removes background from webcam without a green screen.
@@ -65,7 +67,7 @@ Check out our other plugins:
 - [Polyglot](https://github.com/occ-ai/obs-polyglot) translation AI plugin for real-time, local translation to hunderds of languages
 - [URL/API Source](https://github.com/occ-ai/obs-urlsource) will connect to any URL/API HTTP and get the data/image/audio to your scene.
 
-If you like this work, which is given to you completely free of charge, please consider supporting it on GitHub: https://github.com/sponsors/royshil
+If you like this work, which is given to you completely free of charge, please consider supporting it https://github.com/sponsors/royshil or https://www.patreon.com/RoyShilkrot
 
 ## Download
 Check out the [latest releases](https://github.com/occ-ai/obs-ocr/releases) for downloads and install instructions.

cmake/BuildInja.cmake

@@ -0,0 +1,10 @@
+set(INJA_USE_EMBEDDED_JSON UNIX AND NOT APPLE)
+set(INJA_INSTALL OFF)
+set(INJA_EXPORT OFF)
+set(BUILD_TESTING OFF)
+set(INJA_BUILD_TESTS OFF)
+set(BUILD_BENCHMARK OFF)
+set(COVERALLS OFF)
+# inja uses a lot of shadowing
+add_compile_options(-Wno-shadow)
+add_subdirectory(${CMAKE_SOURCE_DIR}/vendor/inja ${CMAKE_BINARY_DIR}/inja EXCLUDE_FROM_ALL)

src/filter-data.h

@@ -38,6 +38,7 @@ struct filter_data {
 	size_t word_length;
 	size_t window_size;
 	uint32_t update_timer_ms;
+	std::string output_format_template;
 
 	bool isDisabled;
 

src/ocr-filter.cpp

@@ -127,6 +127,10 @@ obs_properties_t *ocr_filter_properties(void *data)
 
 	obs_property_set_modified_callback(enable_smoothing_property, enable_smoothing_modified);
 
+	// Output formatting
+	obs_properties_add_text(props, "output_formatting", obs_module_text("OutputFormatting"),
+				OBS_TEXT_MULTILINE);
+
 	// Add a property for the output text source
 	obs_property_t *sources =
 		obs_properties_add_list(props, "text_sources", obs_module_text("OutputTextSource"),
@@ -162,6 +166,7 @@ void ocr_filter_defaults(obs_data_t *settings)
 	obs_data_set_default_bool(settings, "enable_smoothing", false);
 	obs_data_set_default_int(settings, "word_length", 5);
 	obs_data_set_default_int(settings, "window_size", 10);
+	obs_data_set_default_string(settings, "output_formatting", "{{output}}");
 }
 
 void ocr_filter_update(void *data, obs_data_t *settings)
@@ -180,6 +185,7 @@ void ocr_filter_update(void *data, obs_data_t *settings)
 	tf->word_length = obs_data_get_int(settings, "word_length");
 	tf->window_size = obs_data_get_int(settings, "window_size");
 	tf->update_timer_ms = (uint32_t)obs_data_get_int(settings, "update_timer");
+	tf->output_format_template = obs_data_get_string(settings, "output_formatting");
 
 	// Initialize the Tesseract OCR model
 	initialize_tesseract_ocr(tf);

src/tesseract-ocr-utils.cpp

@@ -8,6 +8,8 @@
 
 #include <tesseract/baseapi.h>
 
+#include <inja/inja.hpp>
+
 #include <string>
 #include <fstream>
 #include <deque>
@@ -181,6 +183,15 @@ std::string CharacterBasedSmoothingFilter::add_reading(const std::string &inWord
 	return smoothed_word;
 }
 
+std::string format_text_with_template(inja::Environment &env, const std::string &text,
+				      struct filter_data *tf)
+{
+	// Replace the {{output}} placeholder with the source text using inja
+	nlohmann::json data;
+	data["output"] = text;
+	return env.render(tf->output_format_template, data);
+}
+
 void stop_and_join_tesseract_thread(struct filter_data *tf)
 {
 	{
@@ -209,6 +220,8 @@ void tesseract_thread(void *data)
 
 	obs_log(LOG_INFO, "Starting Tesseract thread, update timer: %d", tf->update_timer_ms);
 
+	inja::Environment env;
+
 	while (true) {
 		{
 			std::lock_guard<std::mutex> lock(tf->tesseract_mutex);
@@ -238,6 +251,7 @@ void tesseract_thread(void *data)
 				if (!ocr_result.empty() &&
 				    is_valid_output_source_name(tf->output_source_name)) {
 					// If an output source is selected - send the results there
+					ocr_result = format_text_with_template(env, ocr_result, tf);
 					setTextCallback(ocr_result, tf);
 				}
 			} catch (const std::exception &e) {