From 64c3f588cbea3bbfd5a23612a4337c773b6a1a38 Mon Sep 17 00:00:00 2001
From: Daniel Cloud <dcloud@sunlightfoundation.com>
Date: Mon, 24 Feb 2014 11:33:52 -0500
Subject: [PATCH] Add core ShapelyFormatter and WKFormatter, which outputs WKT
 which may or not be valid. So #2 is far along now.

---
 hocrgeo/formatters/__init__.py |  2 ++
 hocrgeo/formatters/core.py     | 42 ++++++++++++++++++++++++++++++++++
 hocrgeo/formatters/wk.py       | 13 +++++++++++
 requirements.txt               |  2 +-
 4 files changed, 58 insertions(+), 1 deletion(-)
 create mode 100644 hocrgeo/formatters/core.py
 create mode 100644 hocrgeo/formatters/wk.py

diff --git a/hocrgeo/formatters/__init__.py b/hocrgeo/formatters/__init__.py
index e69de29..eb81445 100644
--- a/hocrgeo/formatters/__init__.py
+++ b/hocrgeo/formatters/__init__.py
@@ -0,0 +1,2 @@
+from hocrgeo.formatters.core import ShapelyFormatter
+from hocrgeo.formatters.wk import WKFormatter
diff --git a/hocrgeo/formatters/core.py b/hocrgeo/formatters/core.py
new file mode 100644
index 0000000..3c5499a
--- /dev/null
+++ b/hocrgeo/formatters/core.py
@@ -0,0 +1,42 @@
+from shapely.geometry import Polygon, box
+
+from shapely import speedups
+
+if speedups.available:
+    speedups.enable()
+
+class ShapelyFormatter:
+    '''Formats a HOCRDocument into shapely geometry'''
+    def __init__(self, hocr_document=None):
+        if hocr_document:
+            self.parse_hocr(hocr_document)
+
+    def parse_hocr(self, hocr_document):
+        '''
+        Parse a hocr_document created by from hocrgeo.parsers.hocr.HOCRParser
+
+        :param hocr_document: instance of hocrparser.document.
+        '''
+
+        def _extract_polys_from_feature_tree(polygons, root, feature_keys):
+            def _poly_from_object(obj):
+                bbox = obj.get('bbox', None)
+                if bbox:
+                    poly = box(bbox.get('x0'), bbox.get('y0'), bbox.get('x1'), bbox.get('y1'))
+                    return poly
+                return None
+
+            features = root.get(feature_keys[0])
+            for f in features:
+                poly = _poly_from_object(f)
+                polygons.append(poly)
+
+                child_keys = feature_keys[1:]
+                if child_keys:
+                    _extract_polys_from_feature_tree(polygons, f, child_keys)
+
+        self._polygons = []
+
+        features = ('pages', 'careas', 'paragraphs', 'lines', 'words')
+
+        _extract_polys_from_feature_tree(self._polygons, hocr_document, features)
diff --git a/hocrgeo/formatters/wk.py b/hocrgeo/formatters/wk.py
new file mode 100644
index 0000000..60f9661
--- /dev/null
+++ b/hocrgeo/formatters/wk.py
@@ -0,0 +1,13 @@
+from shapely.geometry import Polygon, box
+
+from hocrgeo.formatters.core import ShapelyFormatter
+
+class WKFormatter(ShapelyFormatter):
+    """Formats a HOCRParser document as well-known text"""
+
+    @property
+    def wkt(self):
+        if self._polygons:
+            return u'\n'.join([p.wkt for p in self._polygons])
+        return None
+
diff --git a/requirements.txt b/requirements.txt
index c6f0a25..bfe5495 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,4 @@
 chardet
 lxml
 beautifulsoup4
-geojson
\ No newline at end of file
+Shapely
\ No newline at end of file