config

git-svn-id: https://projectname.googlecode.com/svn/trunk@19 c416075f-80b4-e980-4839-00ea3ed24e77

trunk/src/default_main.cpp

@@ -43,15 +43,15 @@ class accumulator: public arr<boost::shared_ptr<T>, boost::shared_ptr<T> > {
     }
 };
 
-void default_main(const program_options& po) {
+void default_main() {
     struct : end< boost::shared_ptr<file_type::base> > {
         void next(const boost::shared_ptr<file_type::base>& t) { std::cout << t->path().string() << "\n"; }
         void stop() { std::cout << "\n"; }
     } output;
 
-    make_pair(po.input_files().begin(), po.input_files().end())
+    make_pair(program_options::input_files().begin(), program_options::input_files().end())
     >>= fs::recursive()
-    >>= (po.extensions().empty() ? The< arr<fs::path, fs::path> >() : The<elem_filter>(po.extensions()))
+    >>= (program_options::extensions().empty() ? The< arr<fs::path, fs::path> >() : The<elem_filter>(program_options::extensions()))
     >>= file_typer_match_first()
     >>= clusterization()
     >>= comparator<file_type::base, file_type::base>()

trunk/src/file_types/base.cpp

@@ -39,4 +39,15 @@ inline comparison_result base::precompare(const boost::shared_ptr<base>& a) cons
     return _size < a->_size ? less : _size == a->_size ? equal : greater;
 }
 
+bool base::check_type(const std::vector<std::string>& types) const {
+    bool res;
+    if (!_mime.empty()) {
+        res = std::find(types.begin(), types.end(), _mime) != types.end();
+    } else {
+        std::string ext = boost::to_lower_copy(_path.extension().string());
+        res = std::find(types.begin(), types.end(), ext) != types.end();
+    }
+    return res;
+}
+
 }

trunk/src/file_types/base.h

@@ -30,17 +30,7 @@ class base {
     virtual boost::shared_ptr<base> compare(const boost::shared_ptr<base>& a) const;
     virtual comparison_result precompare(const boost::shared_ptr<base>& a) const;
 
-    template<typename It1, typename It2>
-    bool check_type(It1 mimes, It1 mimes_end, It2 exts, It2 exts_end) const {
-        bool res;
-        if (!_mime.empty()) {
-            res = std::find(mimes, mimes_end, _mime) != mimes_end;
-        } else {
-            std::string ext = boost::to_lower_copy(_path.extension().string());
-            res = std::find(exts, exts_end, ext) != exts_end;
-        }
-        return res;
-    }
+    bool check_type(const std::vector<std::string>& types) const;
 };
 
 }

trunk/src/file_types/img.cpp

@@ -1,25 +1,26 @@
 #include "img.h"
 #include "../kleisli.h"
 #include "../logger.h"
+#include "../program_options.h"
 
 namespace file_type {
 
-static const unsigned int PIX_SIZE = 128;
-static const double THRESHOLD = 0.005;
 static const unsigned int PIX_IN_INT = sizeof(unsigned int) >= sizeof(Magick::PixelPacket)
     ? sizeof(unsigned int) / sizeof(Magick::PixelPacket) : 1;
 static const float EPSILON = 0.00001;
-static const unsigned int MAX_DIFF = 200000;
 
 img::img(const fs::path& p, const Magick::Image& image): base(p), _pixels(0) {
     const Magick::PixelPacket* pix = image.getConstPixels(0, 0, image.size().width(), image.size().height());
-    static const unsigned int BUCKET_SIZE = 65536 / BUCKET_COUNT;
-    for (unsigned int j = 0; j < BUCKET_COUNT; ++j) {
-        bucket[0][j] = bucket[1][j] = bucket[2][j] = 0;
+    static const unsigned int BUCKET_SIZE = 65536 / program_options::image_bucket_count();
+    for (unsigned int j = 0; j < program_options::image_bucket_count(); ++j) {
+        bucket[0].push_back(0);
+        bucket[1].push_back(0);
+        bucket[2].push_back(0);
     }
     unsigned int size = image.size().width() * image.size().height();
     for (unsigned int i = 0; i < size; ++i, ++pix) {
-        for (unsigned int j = 0, c = BUCKET_SIZE; j < BUCKET_COUNT; ++j, c += BUCKET_SIZE) {
+        for (unsigned int j = 0, c = BUCKET_SIZE;
+            j < program_options::image_bucket_count(); ++j, c += BUCKET_SIZE) {
             if (pix->red < c) {
                 ++bucket[0][j];
             }
@@ -31,7 +32,7 @@ img::img(const fs::path& p, const Magick::Image& image): base(p), _pixels(0) {
             }
         }
     }
-    for (unsigned int j = 0; j < BUCKET_COUNT; ++j) {
+    for (unsigned int j = 0; j < program_options::image_bucket_count(); ++j) {
         for (unsigned int k = 0; k < HISTOGRAM_COUNT; ++k) {
             bucket[k][j] /= size;
         }
@@ -45,10 +46,7 @@ img::~img() {
 }
 
 boost::shared_ptr<img> img::try_file(const boost::shared_ptr<base>& file) {
-    static const std::string mimes[] = { "image/jpeg", "image/png" };
-    static const std::string exts[] = { ".jpg", ".png" };
-    if (file->check_type(mimes, mimes + sizeof(mimes)/sizeof(std::string),
-      exts, exts + sizeof(exts)/sizeof(std::string))) {
+    if (file->check_type(program_options::image_formats())) {
         try { 
             Magick::Image image;
             image.read(file->path().string()); 
@@ -65,15 +63,17 @@ static double abs(double x) { return x >= 0 ? x : -x; }
 boost::shared_ptr<base> img::compare(const boost::shared_ptr<base>& _a) const {
     const img* a = static_cast<const img*>(_a.get());
     {   double res = 0;
-        for (unsigned int i = 0; i < BUCKET_COUNT; ++i) {
+        for (unsigned int i = 0; i < program_options::image_bucket_count(); ++i) {
             for (unsigned int k = 0; k < HISTOGRAM_COUNT; ++k) {
                 res += abs(bucket[k][i] - a->bucket[k][i]);
             }
         }
-        if (res > THRESHOLD) {
+        if (res > program_options::image_threshold()) {
             return boost::shared_ptr<img>();
         }
-    } { init_pixels();
+    }
+    if (program_options::image_precise()) {
+        init_pixels();
         a->init_pixels();
         unsigned int res = 0;
         const unsigned int width0 = width(), width1 = a->width();
@@ -88,7 +88,7 @@ boost::shared_ptr<base> img::compare(const boost::shared_ptr<base>& _a) const {
                 res += abs(_pixels[p].red - a->_pixels[p1].red)
                     + abs(_pixels[p].green - a->_pixels[p1].green)
                     + abs(_pixels[p].blue - a->_pixels[p1].blue);
-                if (res > MAX_DIFF) {
+                if (res > program_options::image_max_diff()) {
                     return boost::shared_ptr<img>();
                 }
             }
@@ -118,7 +118,8 @@ void img::init_pixels() const {
     if (_pixels == 0) {
         Magick::Image im;
         im.read(path().string()); 
-        #define CALC_SIZE(x) ((x) >= PIX_SIZE ? PIX_SIZE : round(x))
+        #define CALC_SIZE(x) ((x) >= program_options::image_img_size() \
+            ? program_options::image_img_size() : round(x))
         im.scale(Magick::Geometry(CALC_SIZE(im.size().width()), CALC_SIZE(im.size().height())));
         const Magick::PixelPacket* pix = im.getConstPixels(0, 0, im.size().width(), im.size().height());
         _pixels = new Magick::PixelPacket[im.size().width() * im.size().height() + 2 * PIX_IN_INT];

trunk/src/file_types/img.h

@@ -1,6 +1,7 @@
 #ifndef _FILE_TYPE_IMG_H_
 #define _FILE_TYPE_IMG_H_
 
+#include <vector>
 #include <boost/shared_ptr.hpp>
 #include <Magick++.h>
 
@@ -16,9 +17,8 @@ struct img: base {
     boost::shared_ptr<base> compare(const boost::shared_ptr<base>& a) const;
     comparison_result precompare(const boost::shared_ptr<base>& a) const;
 private:
-    static const unsigned int BUCKET_COUNT = 4;
     static const unsigned int HISTOGRAM_COUNT = 3;
-    double bucket[HISTOGRAM_COUNT][BUCKET_COUNT];
+    std::vector<double> bucket[HISTOGRAM_COUNT];
     mutable Magick::PixelPacket* _pixels;
     float _aspect_ratio;
     void init_pixels() const;

trunk/src/file_types/text.cpp

@@ -2,11 +2,11 @@
 #include <boost/make_shared.hpp>
 #include <string.h>
 #include <ctype.h>
-#include <algorithm>
 #include <map>
 
 #include "text.h"
 #include "../kleisli.h"
+#include "../program_options.h"
 
 namespace file_type {
 
@@ -88,74 +88,63 @@ text::text(const fs::path& p): base(p) {
     char buf[buf_size];
     fs::ifstream file(p);
     uint32_t hash = hash_init();
-    std::string last;
-    std::map<std::string, unsigned int> counts;
+    unsigned int cur_size = 0;
+    std::map<uint32_t, unsigned int> counts;
     while (!file.eof()) {
         file.read(buf, buf_size);
         for (int i = 0; i < file.gcount(); ++i) {
-            if (buf[i] == '.' || buf[i] == '!' || buf[i] == '?') {
-                _hashes.push_back(hash_final(hash));
-                hash = hash_init();
-            } else
             if (isalnum(buf[i])) {
                 hash_update(hash, tolower(buf[i]));
-            }
-            if (isalnum(buf[i])) {
-                last += /*tolower*/(buf[i]);
+                ++cur_size;
             } else {
-                if (last.length() > 3) {
-                    ++counts[last];
-                    last = "";
+                if (cur_size > 3) {
+                    ++counts[hash_final(hash)];
                 }
+                hash = hash_init();
+                cur_size = 0;
             }
         }
     }
-    std::map<unsigned int, std::vector<std::string> > rev;
-    for (std::map<std::string, unsigned int>::iterator it = counts.begin(); it != counts.end(); ++it) {
+    std::map<unsigned int, std::vector<uint32_t> > rev;
+    for (std::map<uint32_t, unsigned int>::iterator it = counts.begin(); it != counts.end(); ++it) {
         rev[it->second].push_back(it->first);
     }
-    std::map<unsigned int, std::vector<std::string> >::iterator it = --rev.end();
-    for (unsigned int i = 0; i < WORDS_COUNT;) {
-        for (unsigned int j = 0; j < it->second.size(); ++j, ++i) {
-            if (i >= WORDS_COUNT) {
+    std::map<unsigned int, std::vector<uint32_t> >::iterator it = --rev.end();
+    for (unsigned int i = 0; i < program_options::text_words_count();) {
+        for (unsigned int j = 0; j < it->second.size(); ++j) {
+            if (i++ >= program_options::text_words_count()) {
                 break;
             }
-            _words[i] = it->second[j];
+            _words.push_back(it->second[j]);
         }
         --it;
     }
-    std::sort(_words, _words + WORDS_COUNT);
-    std::sort(_hashes.begin(), _hashes.end());
+    std::sort(_words.begin(), _words.end());
 }
 
 boost::shared_ptr<text> text::try_file(const boost::shared_ptr<base>& file) {
-    static const std::string mimes[] = { "text/plain" };
-    static const std::string exts[] = { ".txt" };
-    return file->check_type(mimes, mimes + sizeof(mimes)/sizeof(std::string),
-        exts, exts + sizeof(exts)/sizeof(std::string)) ?
+    return file->check_type(program_options::text_formats()) ?
         boost::make_shared<text>(file->path()) : boost::shared_ptr<text>();
 }
 
 boost::shared_ptr<base> text::compare(const boost::shared_ptr<base>& _a) const {
     const text* a = static_cast<const text*>(_a.get());
     unsigned int r = 0;
-    const std::string *first1 = _words, *last1 = _words + WORDS_COUNT,
-        *first2 = a->_words, *last2 = a->_words + WORDS_COUNT;
-    // std::vector<uint32_t>::const_iterator first1 = _hashes.begin(),
-    //     last1 = _hashes.end(), first2 = a->_hashes.begin(), last2 = a->_hashes.end();
+    std::vector<uint32_t>::const_iterator
+        first1 = _words.begin(), last1 = _words.end(),
+        first2 = a->_words.begin(), last2 = a->_words.end();
     while (true) {
         if (first1 == last1)
             { r += distance(first2, last2); break; }
         if (first2 == last2)
             { r += distance(first1, last1); break; }
         if (*first1 < *first2)
-            { ++r; *first1++; }
+            { ++r; first1++; }
         else if (*first2 < *first1)
-            { ++r; *first2++; }
+            { ++r; first2++; }
         else { first1++; first2++; }
     }
-    return r <= 2 ? _a : boost::shared_ptr<text>();
-    // return float(r) / (_hashes.size() + a->_hashes.size()) < THRESHOLD ? _a : boost::shared_ptr<text>();
+    return r <= program_options::text_threshold() ? _a : boost::shared_ptr<text>();
 }
 
 }

trunk/src/file_types/text.h

@@ -15,9 +15,7 @@ struct text: base {
     boost::shared_ptr<base> compare(const boost::shared_ptr<base>& a) const;
     comparison_result precompare(const boost::shared_ptr<base>& a) const { return equal; }
 private:
-    static const unsigned int WORDS_COUNT = 10;
-    std::vector<uint32_t> _hashes;
-    std::string _words[WORDS_COUNT];
+    std::vector<uint32_t> _words;
 };
 
 }

trunk/src/main.cpp

@@ -2,17 +2,17 @@
 #include "program_options.h"
 #include "magic_file.h"
 
-void default_main(const program_options& po);
+void default_main();
 
 int main(int argc, char* argv[]) {
     stderr_logger std_logger(argv[0]);
     logger::set_std(&std_logger);
 
-    program_options po(argc, argv);
+    program_options::initialize(argc, argv);
 
     magic::initialize();
 
-    default_main(po);
+    default_main();
 
     magic::destroy();