gocr-ruby 0.0.1 → 0.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 18b9142ede00b0c09bc41a71caa18a5090ad44e2
-  data.tar.gz: b72f4aed15d04583f8330f0f3bccebc0a9649f9e
+  metadata.gz: 1d257eb31f2bf08f056a30e19597da354ced4147
+  data.tar.gz: b4ed2b93a9fd6aeedf31e874bfec01b14f40b56c
 SHA512:
-  metadata.gz: 8ae9ddb96adb3ddf550e59347f597927df70e17df633e9bbb24e0d45b534bdd43bb5075dd3ada115fea4343cefe0f67ccdb53c209a5dd94d925a537ffd03b004
-  data.tar.gz: 260c67406fb4af5240939c6b16aea210667c907c0ee36ed13a7b94b691485c2ddfada467eb82ea73a4b3779a68e82213f035f0df40ef293177921f08e634771a
+  metadata.gz: 8df4f786fbc95987f1cc37c722af93e768d3c71fd53ff108a23184aca8bfe2e1cb1c1b37d0647785ac1267d09ca3d863ea8dba7d9c3e6673b851808e6611622f
+  data.tar.gz: b7579da8c2e8f2fa1e385c0afbf484a105ecf7368b0727a55cdfff31c9451ef2d9223886a18580c6af647a53a621a5b4496a2147f851acd025699233e13223a0

data/.gitignore

@@ -19,4 +19,6 @@ tmp
 *.so
 .rvmrc
 .idea/
-*.log
+*.log
+
+GOCR-Ruby.*

data/README.md

@@ -1,6 +1,8 @@
 # gocr-ruby is a gocr binding for ruby
 
-TODO: Write a gem description
+Ruby GOCR binding for lib http://jocr.sourceforge.net/
+
+Currently it is in progress, unstable version.
 
 ## Installation
 
@@ -18,7 +20,25 @@ Or install it yourself as:
 
 ## Usage
 
-TODO: Write usage instructions here
+current api almost as a ruby-tesseract-ocr
+
+```ruby
+require 'gocr'
+e = GOCR::Engine.new(whitelist: '0-9').text_for('image.png')
+```
+
+### Available options
+
+* `:whitelist` - char filter (ex. hexdigits: ""0-9A-Fx"", only ASCII)
+* `:database` - database path including final slash (default is ./db/)
+* `:format` - output format (ISO8859_1 TeX HTML XML UTF8 ASCII)
+* `:gray_level` - threshold grey level 0<160<=255 (0 = autodetect)
+* `:numbers_only` - numbers only
+* `:mode` operation modes (bitpattern, see official gocr manual)
+* `:certainty` - value of certainty (in percent, 0..100, default=95)
+* `:unrecognize_char` - output this string for every unrecognized character
+* `:dust_size` - dust_size (remove small clusters, -1 = autodetect)
+* `:space_width` - spacewidth/dots (0 = autodetect)
 
 ## Contributing
 

data/Rakefile

@@ -9,7 +9,7 @@ require 'rubygems/package_task'
 #
 # See https://github.com/luislavena/rake-compiler for details
 
-Rake::ExtensionTask.new 'gocr-ruby' do |ext|
+Rake::ExtensionTask.new 'gocr' do |ext|
 
   # This causes the shared object to be placed in lib/my_malloc/my_malloc.so
   #
@@ -19,7 +19,7 @@ Rake::ExtensionTask.new 'gocr-ruby' do |ext|
   ext.lib_dir = 'lib/gocr'
 end
 
-s = Gem::Specification.new 'gocr-ruby', '0.0.1' do |s|
+s = Gem::Specification.new 'gocr', '0.0.1' do |s|
   s.summary = 'simple gocr wrapper'
   s.authors = %w[zyablitskiy@gmail.com]
 
@@ -40,7 +40,7 @@ Gem::PackageTask.new s do end
 # This isn't a good test, but does provide a sanity check
 
 task test: %w[compile] do
-  ruby '-Ilib', '-rgocr', '-e', 'p GOCR::Image.recognize("image.png")'
+  ruby '-Ilib', '-rgocr', '-e', 'p GOCR::Engine.new(numbers_only: false).text_for("image.png")'
 #  ruby '-Ilib', '-rgocr', '-e', 'p 3'
 end
 

data/ext/gocr/gocr.c

@@ -42,7 +42,7 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
 #include "ocr0.h"		/* only_numbers */
 #include "progress.h"
 #include "version.h"
-#include "ruby.h"
+#include <ruby.h>
 
 /*
 #ifndef RSTRING_PTR
@@ -322,9 +322,12 @@ static int read_picture(job_t *job) {
 }
 
 /* subject of change, we need more output for XML (ToDo) */
-void print_output(job_t *job) {
+char* print_output(job_t *job) {
   int linecounter = 0;
   const char *line;
+  char** lines;
+  char* output;
+  int i = 0, j, n = 0;
 
   assert(job);
         
@@ -332,61 +335,131 @@ void print_output(job_t *job) {
    simplify code 2010-09-26
 */
   linecounter = 0;
+  lines = (char**) malloc(job->res.lines.num);
   line = getTextLine(&(job->res.linelist), linecounter++);
   while (line) {
+    n += strlen(line) + 1;
+    lines[i] = (char*) malloc(strlen(line));
+    strcpy(lines[i++], line);
     /* notice: decode() is shiftet to getTextLine since 0.38 */
-    fputs(line, stdout);
     if (job->cfg.out_format==HTML) fputs("<br />",stdout);
     if (job->cfg.out_format!=XML)  fputc('\n', stdout);
     line = getTextLine(&(job->res.linelist), linecounter++);
   }
-  free_textlines(&(job->res.linelist));
+//  free_textlines(&(job->res.linelist));
+
+  output = (char*) malloc(n);
+  strcpy(output, lines[0]);
+  for(j = 1; j < i; j++) {
+      strcat(output, "\n");
+      strcat(output, lines[j]);
+  }
+
+  return output;
 }
 
 /* FIXME jb: remove JOB; renamed to OCR_JOB 2010-09-26 */
 job_t *OCR_JOB;
 
 
-char* gocr_recognize(char* filename) {
-    char* line;
-    int multipnm=1;
-    job_t job1, *job; /* fixme, dont want global variables for lib */
-    job=OCR_JOB=&job1;
-
-    setvbuf(stdout, (char *) NULL, _IONBF, 0);	/* not buffered */
+char* gocr_main(job_t* job) {
+  int multipnm=1;
+  char* output;
+  setvbuf(stdout, (char *) NULL, _IONBF, 0);	/* not buffered */
 
-    job_init(job); /* init cfg and db */
-    job->src.fname = filename;
-    /* load character data base (JS1002: now outside pgm2asc) */
-    if ( job->cfg.mode & 2 ) /* check for db-option flag */
-        load_db(job);
+  /* load character data base (JS1002: now outside pgm2asc) */
+  if ( job->cfg.mode & 2 ) /* check for db-option flag */
+    load_db(job);
     /* load_db uses readpnm() and would conflict with multi images */
 
+  while (multipnm==1) { /* multi-image loop */
+
     job_init_image(job); /* single image */
+
     mark_start(job);
-    read_picture(job);
 
+    multipnm = read_picture(job);
+    /* separation of main and rest for using as lib
+       this will be changed later => introduction of set_option()
+       for better communication to the engine  */
+    if (multipnm<0) break; /* read error */
+
+    /* call main loop */
     pgm2asc(job);
+
     mark_end(job);
-    line = getTextLine(&(job->res.linelist), 0);
+
+    output = print_output(job);
+
     job_free_image(job);
-    return line;
+
+  }
+
+//  return ((multipnm<0)?multipnm:0); /* -1=255 on error, 0 ok */
+    return output;
 }
 
 
 static VALUE image_recognize(VALUE self, VALUE arg) {
-  char* filename = StringValuePtr(arg);
-  return rb_str_new2( gocr_recognize(filename) );
+    VALUE tmp;
+
+    job_t job1, *job; /* fixme, dont want global variables for lib */
+    job=OCR_JOB=&job1;
+    job_init(job);
+    
+    job->src.fname = StringValuePtr(arg);
+    tmp = rb_iv_get(self, "@database");
+    if (tmp != Qnil) {
+        job->cfg.db_path = StringValuePtr(tmp);
+    }
+    tmp = rb_iv_get(self, "@format");
+    if (tmp != Qnil) {
+        job->cfg.out_format = NUM2INT(tmp);
+    }
+    tmp = rb_iv_get(self, "@whitelist");
+    if (tmp != Qnil) {
+        if (strlen(StringValuePtr(tmp)) > 0)
+            job->cfg.cfilter = StringValuePtr(tmp);
+    }
+    tmp = rb_iv_get(self, "@dust_size");
+    if (tmp != Qnil) {
+        job->cfg.dust_size = NUM2INT(tmp);
+    }
+    tmp = rb_iv_get(self, "@gray_level");
+    if (tmp != Qnil) {
+        job->cfg.cs = NUM2INT(tmp);
+    }
+    tmp = rb_iv_get(self, "@space_width");
+    if (tmp != Qnil) {
+        job->cfg.spc = NUM2INT(tmp);
+    }
+    tmp = rb_iv_get(self, "@mode");
+    if (tmp != Qnil) {
+        job->cfg.mode |= NUM2INT(tmp);
+    }
+    tmp = rb_iv_get(self, "@numbers_only");
+    if (tmp == Qtrue) {
+        job->cfg.only_numbers = 1;
+    }
+    tmp = rb_iv_get(self, "@certainty");
+    if (tmp != Qnil) {
+        job->cfg.certainty = NUM2INT(tmp);
+    }
+    tmp = rb_iv_get(self, "@unrecognize_char");
+    if (tmp != Qnil) {
+        job->cfg.unrec_marker = StringValuePtr(tmp)[0];
+    }
+    return rb_str_new2( gocr_main(job) );
 }
 
 /*
- * @brief define ruby class GOCR::Image and method recognize
+ * @brief define ruby class GOCR::Engine and method text_for
  *
  */
 void Init_gocr() {
   VALUE mGocr = rb_define_module("GOCR");
-  VALUE mImage = rb_define_class_under(mGocr, "Image", rb_cObject);
-  rb_define_singleton_method(mImage, "recognize", image_recognize, 1);
+  VALUE mImage = rb_define_class_under(mGocr, "Engine", rb_cObject);
+  rb_define_method(mImage, "text_for", image_recognize, 1);
 }
 
 
@@ -433,4 +506,4 @@ void Init_gocr() {
 //  }
 //
 //  return ((multipnm<0)?multipnm:0); /* -1=255 on error, 0 ok */
-//}
+//}

data/ext/gocr/unicode.h

@@ -44,7 +44,7 @@ Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
 #include <stddef.h>
 
 enum format {
-	ISO8859_1, TeX, HTML, XML, SGML, UTF8, ASCII
+    UTF8, ISO8859_1, TeX, HTML, XML, SGML, ASCII
 };
 typedef enum format FORMAT;
 

data/lib/gocr.rb

@@ -1,5 +1,5 @@
 require "gocr/version"
-require "gocr/image"
+require "gocr/engine"
 
 module GOCR
   # Your code goes here...

data/lib/gocr/engine.rb

@@ -0,0 +1,20 @@
+require "gocr/gocr"
+
+module GOCR
+  class Engine
+    attr_accessor :whitelist, :blacklist, :database, :format, :gray_level, :numbers_only,
+                  :mode, :certainty, :unrecognize_char, :dust_size, :space_width
+
+    FORMATS = Hash[%w(UTF8 ISO8859_1 TeX HTML XML ASCII).map.with_index.to_a].freeze
+
+    def initialize(options={})
+      options.each do |k, v|
+        send("#{k}=", v) if respond_to?(k)
+      end
+      @format = FORMATS[format].to_i
+      @dust_size = -1 if dust_size.nil?
+      @unrecognize_char = @unrecognize_char[0] unless unrecognize_char.nil?
+    end
+
+  end
+end

data/lib/gocr/version.rb

@@ -1,3 +1,3 @@
 module GOCR
-  VERSION = "0.0.1"
+  VERSION = "0.0.2"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: gocr-ruby
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors:
 - Vladimir Zyablitskiy
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-03-28 00:00:00.000000000 Z
+date: 2014-03-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -126,7 +126,7 @@ files:
 - gocr-ruby.gemspec
 - image.png
 - lib/gocr.rb
-- lib/gocr/image.rb
+- lib/gocr/engine.rb
 - lib/gocr/version.rb
 homepage: https://github.com/rainlabs/gocr-ruby
 licenses:

data/lib/gocr/image.rb

@@ -1,8 +0,0 @@
-require "gocr/gocr"
-
-module GOCR
-  class Image
-
-  end
-end
-