[ocrfeeder] Add PDF exportation

[Joaquim Manuel Pereira Rocha <jrocha src gnome org>]

commit d73ebc447451da8816a80ad14e316ab9d6cdda06
Author: Joaquim Rocha <jrocha igalia com>
Date:   Wed Jul 27 15:33:01 2011 +0200

    Add PDF exportation
    
    Exports to a PDF from scratch. Might not respect some of
    configurations like the fonts and letter spacing which should be
    fine-tuned in the future.

 src/ocrfeeder/feeder/documentGeneration.py |   47 ++++++++++++++++++++++++++++
 src/ocrfeeder/studio/studioBuilder.py      |    6 +++-
 src/ocrfeeder/studio/widgetModeler.py      |   14 ++++++++-
 3 files changed, 65 insertions(+), 2 deletions(-)
---
diff --git a/src/ocrfeeder/feeder/documentGeneration.py b/src/ocrfeeder/feeder/documentGeneration.py
index ba218aa..3b2b2be 100644
--- a/src/ocrfeeder/feeder/documentGeneration.py
+++ b/src/ocrfeeder/feeder/documentGeneration.py
@@ -29,6 +29,9 @@ from ocrfeeder.util import TEXT_TYPE, IMAGE_TYPE, ALIGN_LEFT, ALIGN_RIGHT, ALIGN
     ALIGN_FILL
 from ocrfeeder.util.graphics import getImagePrintSize
 from ocrfeeder.util.lib import debug
+from reportlab.pdfgen import canvas
+from reportlab.lib import units
+from reportlab.lib.utils import ImageReader
 import math
 import os.path
 import shutil
@@ -349,3 +352,47 @@ class PlaintextGenerator(DocumentGenerator):
                 f.close() # Close the file
         except IOError:
             pass
+
+class PdfGenerator(DocumentGenerator):
+
+    def __init__(self, name):
+        self.name = name
+        self.canvas = canvas.Canvas(self.name)
+        self.page_data = None
+
+    def addText(self, box):
+        x, y, width, height = box.getBoundsPrintSize(self.page_data.resolution)
+        text = self.canvas.beginText()
+        text.setTextOrigin(x * units.inch,
+                           (self.page_data.height - y) * units.inch)
+        text.setCharSpace(box.text_data.letter_space)
+        text.setLeading(box.text_data.line_space + box.text_data.size)
+        text.moveCursor(0, box.text_data.size)
+        try:
+            self.setFont(box.text_data.face,
+                         box.text_data.size)
+        except:
+            debug('Error setting font %s' % box.text_data.face)
+            self.canvas.setFontSize(box.text_data.size)
+        for line in box.text.split('\n'):
+            text.textLine(line)
+        self.canvas.drawText(text)
+
+    def addImage(self, box):
+        x, y, width, height = box.getBoundsPrintSize(self.page_data.resolution)
+        self.canvas.drawInlineImage(box.image,
+                                    x * units.inch,
+                                    (self.page_data.height - (y + height)) * \
+                                        units.inch,
+                                    width * units.inch,
+                                    height * units.inch)
+
+    def addPage(self, page_data):
+        self.canvas.setPageSize((page_data.width * units.inch,
+                                 page_data.height * units.inch))
+        self.page_data = page_data
+        self.addBoxes(page_data.data_boxes)
+        self.canvas.showPage()
+
+    def save(self):
+        self.canvas.save()
diff --git a/src/ocrfeeder/studio/studioBuilder.py b/src/ocrfeeder/studio/studioBuilder.py
index 0f0599e..33d34c2 100644
--- a/src/ocrfeeder/studio/studioBuilder.py
+++ b/src/ocrfeeder/studio/studioBuilder.py
@@ -60,7 +60,8 @@ class Studio:
 
         self.EXPORT_FORMATS = {0: (self.exportToHtml, _('HTML')),
                                1: (self.exportToOdt, _('ODT')),
-                               2: (self.exportToPlaintext, _('Plain Text'))}
+                               2: (self.exportToPdf, _('PDF')),
+                               3: (self.exportToPlaintext, _('Plain Text'))}
 
         self.title = OCRFEEDER_STUDIO_NAME
         self.main_window = widgetPresenter.MainWindow()
@@ -290,6 +291,9 @@ class Studio:
     def exportToPlaintext(self, widget = None):
         self.source_images_controler.exportPagesToPlaintext(self.source_images_selector.getPixbufsSorted())
 
+    def exportToPdf(self, widget = None):
+        self.source_images_controler.exportPagesToPdf(self.source_images_selector.getPixbufsSorted())
+
     def exportDialog(self, widget):
         format_names = [format[1] for format in self.EXPORT_FORMATS.values()]
         export_dialog = widgetPresenter.ExportDialog(_('Export pages'), format_names)
diff --git a/src/ocrfeeder/studio/widgetModeler.py b/src/ocrfeeder/studio/widgetModeler.py
index 1ec4b69..0adb850 100644
--- a/src/ocrfeeder/studio/widgetModeler.py
+++ b/src/ocrfeeder/studio/widgetModeler.py
@@ -20,7 +20,7 @@
 
 from customWidgets import SelectableBoxesArea
 from dataHolder import DataBox, PageData, TEXT_TYPE, IMAGE_TYPE
-from ocrfeeder.feeder.documentGeneration import OdtGenerator, HtmlGenerator, PlaintextGenerator
+from ocrfeeder.feeder.documentGeneration import OdtGenerator, HtmlGenerator, PlaintextGenerator, PdfGenerator
 from ocrfeeder.feeder.imageManipulation import *
 from ocrfeeder.feeder.layoutAnalysis import *
 from pango import FontDescription, SCALE
@@ -709,6 +709,18 @@ class ImageReviewer_Controler:
                 document_generator.addText(image_reviewer.getAllText())
             document_generator.save()
 
+    def exportPagesToPdf(self, pixbufs_sorted = []):
+        image_reviewers = self.__askForNumberOfPages(_('Export to PDF'),
+                                                     pixbufs_sorted)
+        if not image_reviewers:
+            return
+        file_name = self.__askForFileName()
+        if file_name:
+            document_generator = PdfGenerator(file_name)
+            for image_reviewer in image_reviewers:
+                document_generator.addPage(image_reviewer.getPageData())
+            document_generator.save()
+
     def saveProjectAs(self):
         return self.__askForFileName(extension = '.ocrf')