Error: {e} Check the server log for more information.'
diff --git a/ocr.py b/ocr.py
index dc21f02..03a4012 100644
--- a/ocr.py
+++ b/ocr.py
@@ -127,6 +127,21 @@ def add_text_document(self, anno):
[text_val] if text_val not in self.text else self.text
+def prepare_ocr(mmif, view, viz_id):
+ """
+ Prepares list of frames that will be passed back and forth between server
+ and client, and renders the first page of the OCR.
+ """
+ ocr_frames = get_ocr_frames(view, mmif)
+
+ # Generate pages (necessary to reduce IO cost) and render
+ frames_list = [(k, vars(v)) for k, v in ocr_frames.items()]
+ frames_list = find_duplicates(frames_list)
+ frames_pages = paginate(frames_list)
+ # Save page list as temp file
+ save_json(frames_pages, view.id, viz_id)
+
+
def find_annotation(anno_id, mmif):
if mmif.id_delimiter in anno_id:
view_id, anno_id = anno_id.split(mmif.id_delimiter)
diff --git a/render.py b/render.py
index 2020550..13c769c 100644
--- a/render.py
+++ b/render.py
@@ -11,7 +11,7 @@
import traceback
from utils import get_status, get_properties, get_abstract_view_type, url2posix, get_vtt_file
-from ocr import get_ocr_frames, paginate, find_duplicates, save_json, make_image_directory, is_duplicate_image
+from ocr import prepare_ocr, make_image_directory, is_duplicate_image
import cv2
import json
import tempfile
@@ -22,7 +22,7 @@
Methods to render MMIF documents and their annotations in various formats.
"""
-# -- Documents --
+# -- Render methods --
def render_documents(mmif, viz_id):
@@ -31,204 +31,263 @@ def render_documents(mmif, viz_id):
"""
tabs = []
for document in mmif.documents:
- try:
- # Add symbolic link to document to static folder, so it can be accessed
- # by the browser.
- doc_path = document.location_path()
- doc_symlink_path = pathlib.Path(
- current_app.static_folder) / cache._CACHE_DIR_SUFFIX / viz_id / (f"{document.id}.{doc_path.split('.')[-1]}")
- os.symlink(doc_path, doc_symlink_path)
- doc_symlink_rel_path = '/' + \
- doc_symlink_path.relative_to(
- current_app.static_folder).as_posix()
+ if document.at_type == DocumentTypes.TextDocument:
+ tabs.append(TextTab(document, viz_id))
+ elif document.at_type == DocumentTypes.ImageDocument:
+ tabs.append(ImageTab(document, viz_id))
+ elif document.at_type == DocumentTypes.AudioDocument:
+ tabs.append(AudioTab(document, viz_id))
+ elif document.at_type == DocumentTypes.VideoDocument:
+ tabs.append(VideoTab(document, mmif, viz_id))
- if document.at_type == DocumentTypes.TextDocument:
- html_tab = render_text(doc_path)
- elif document.at_type == DocumentTypes.ImageDocument:
- html_tab = render_image(doc_symlink_rel_path)
- elif document.at_type == DocumentTypes.AudioDocument:
- html_tab = render_audio(doc_symlink_rel_path)
- elif document.at_type == DocumentTypes.VideoDocument:
- html_tab = render_video(doc_symlink_rel_path, mmif, viz_id)
-
- tabs.append({"id": document.id,
- "tab_name": document.at_type.shortname,
- "html": html_tab})
-
- except Exception:
- tabs.append({"id": document.id,
- "tab_name": document.at_type.shortname,
- "html": f"Error rendering document:
{traceback.format_exc()}"})
return tabs
-def render_text(text_path):
- """Return the content of the text document, but with some HTML tags added."""
- text_path = url2posix(text_path)
- if not os.path.isfile(text_path):
- raise FileNotFoundError(f"File not found: {text_path}")
- with open(text_path) as t_file:
- content = t_file.read().replace("\n", "{traceback.format_exc()}"})
+ abstract_view_type = get_abstract_view_type(view, mmif)
+ if abstract_view_type == "NER":
+ tabs.append(NERTab(mmif, view))
+ elif abstract_view_type == "ASR":
+ tabs.append(VTTTab(mmif, view, viz_id))
+ elif abstract_view_type == "OCR":
+ tabs.append(OCRTab(mmif, view, viz_id))
return tabs
-def render_info(mmif):
- s = StringIO('Howdy')
- s.write("")
- for document in mmif.documents:
- at_type = document.at_type.shortname
- location = document.location
- s.write("%s %s\n" % (at_type, location))
- s.write('\n')
- for view in mmif.views:
- app = view.metadata.app
- status = get_status(view)
- s.write('%s %s %s %d\n' %
- (view.id, app, status, len(view.annotations)))
- if len(view.annotations) > 0:
- s.write('\n')
- types = Counter([a.at_type.shortname
- for a in view.annotations])
- for attype, count in types.items():
- s.write(' %4d %s\n' % (count, attype))
- s.write('\n')
- s.write("")
- return s.getvalue()
+# -- Base Tab Class --
+class DocumentTab():
+ def __init__(self, document, viz_id):
+ self.id = document.id
+ self.tab_name = document.at_type.shortname
+ self.viz_id = viz_id
-def render_annotation_table(mmif):
- s = StringIO('Howdy')
- for view in mmif.views:
- status = get_status(view)
- s.write('%s %s %s %d annotations
\n' - % (view.id, view.metadata.app, status, len(view.annotations))) - s.write("\n") - s.write("\n") - return s.getvalue() - - -def render_jstree(mmif): - return render_template('interactive.html', mmif=mmif, aligned_views=[]) - - -def render_asr_vtt(view, viz_id): - vtt_filename = get_vtt_file(view, viz_id) - with open(vtt_filename) as vtt_file: - vtt_content = vtt_file.read() - return f"\n") - def limit_len(str): return str[:500] + \ - " . . . }" if len(str) > 500 else str - for annotation in view.annotations: - s.write('
\n") - s.write("\n') - s.write(' \n') - s.write("%s \n' % annotation.id) - s.write('%s \n' % annotation.at_type.shortname) - s.write('%s \n' % - limit_len(get_properties(annotation))) - s.write('
{vtt_content}"
-
-
-def render_ner(mmif, view):
- metadata = view.metadata.contains.get(Uri.NE)
- ner_document = metadata.get('document')
- return displacy.visualize_ner(mmif, view, ner_document, current_app.root_path)
-
-
-def prepare_and_render_ocr(mmif, view, viz_id):
- """
- Prepares list of frames that will be passed back and forth between server
- and client, and renders the first page of the OCR.
- """
- vid_path = mmif.get_documents_by_type(DocumentTypes.VideoDocument)[
- 0].location_path()
+ try:
+ # Add symbolic link to document to static folder, so it can be accessed
+ # by the browser.
+ self.doc_path = document.location_path()
+ self.doc_symlink_path = pathlib.Path(
+ current_app.static_folder) / cache._CACHE_DIR_SUFFIX / viz_id / (f"{document.id}.{self.doc_path.split('.')[-1]}")
+ os.symlink(self.doc_path, self.doc_symlink_path)
+ self.doc_symlink_rel_path = '/' + \
+ self.doc_symlink_path.relative_to(
+ current_app.static_folder).as_posix()
+
+ self.html = self.render()
- ocr_frames = get_ocr_frames(view, mmif)
+ except Exception as e:
+ self.html = f"Error rendering document: {traceback.format_exc()}"
+
+ def __str__(self):
+ return f"Tab: {self.tab_name} ({self.id})"
+
+
+class AnnotationTab():
+ def __init__(self, mmif, view=None):
+ self.mmif = mmif
+ # Some AnnotationTab sub-classes don't refer to a specific view, and so
+ # they specify their own ids and tab names. For ones that do refer to
+ # a specific view, we set the ids/tab names based on view properties.
+ if view:
+ self.view = view
+ # Workaround to deal with the fact that some apps have a version number
+ # in the URL
+ app_url = view.metadata.app if re.search(
+ r"\/v\d+\.?\d?$", view.metadata.app) else view.metadata.app + "/v1"
+ app_shortname = app_url.split("/")[-2]
- # Generate pages (necessary to reduce IO cost) and render
- frames_list = [(k, vars(v)) for k, v in ocr_frames.items()]
- frames_list = find_duplicates(frames_list)
- frames_pages = paginate(frames_list)
- # Save page list as temp file
- save_json(frames_pages, view.id, viz_id)
- return render_ocr_page(viz_id, vid_path, view.id, 0)
+ self.id = view.id
+ self.tab_name = f"{app_shortname}-{view.id}"
+ try:
+ self.html = self.render()
+ except Exception as e:
+ self.html = f"Error rendering view: {traceback.format_exc()}"
+
+
+# -- Document Classes --
+
+class TextTab(DocumentTab):
+ def __init__(self, document, viz_id):
+ super().__init__(document, viz_id)
+
+ def render(self):
+ with open(self.doc_path) as t_file:
+ content = t_file.read().replace("\n", "")
+ for document in mmif.documents:
+ at_type = document.at_type.shortname
+ location = document.location
+ s.write("%s %s\n" % (at_type, location))
+ s.write('\n')
+ for view in mmif.views:
+ app = view.metadata.app
+ status = get_status(view)
+ s.write('%s %s %s %d\n' %
+ (view.id, app, status, len(view.annotations)))
+ if len(view.annotations) > 0:
+ s.write('\n')
+ types = Counter([a.at_type.shortname
+ for a in view.annotations])
+ for attype, count in types.items():
+ s.write(' %4d %s\n' % (count, attype))
+ s.write('\n')
+ s.write("")
+ return s.getvalue()
+
+
+class AnnotationTableTab(AnnotationTab):
+ def __init__(self, mmif):
+ self.id = "annotations"
+ self.tab_name = "Annotations"
+ super().__init__(mmif)
+
+ def render(self):
+ mmif = self.mmif
+ s = StringIO('Howdy')
+ for view in mmif.views:
+ status = get_status(view)
+ s.write('%s %s %s %d annotations
\n' + % (view.id, view.metadata.app, status, len(view.annotations))) + s.write("\n") + s.write("\n") + return s.getvalue() + + +class JSTreeTab(AnnotationTab): + def __init__(self, mmif): + self.id = "tree" + self.tab_name = "Tree" + super().__init__(mmif) + + def render(self): + mmif = self.mmif + return render_template('interactive.html', mmif=mmif, aligned_views=[]) + + +class NERTab(AnnotationTab): + def __init__(self, mmif, view): + super().__init__(mmif, view) + + def render(self): + metadata = self.view.metadata.contains.get(Uri.NE) + ner_document = metadata.get('document') + return displacy.visualize_ner(self.mmif, self.view, ner_document, current_app.root_path) + + +class VTTTab(AnnotationTab): + def __init__(self, mmif, view, viz_id): + self.viz_id = viz_id + super().__init__(mmif, view) + + def render(self): + vtt_filename = get_vtt_file(self.view, self.viz_id) + with open(vtt_filename) as vtt_file: + vtt_content = vtt_file.read() + return f"\n") + def limit_len(str): return str[:500] + \ + " . . . }" if len(str) > 500 else str + for annotation in view.annotations: + s.write('
\n") + s.write("\n') + s.write(' \n') + s.write("%s \n' % annotation.id) + s.write('%s \n' % annotation.at_type.shortname) + s.write('%s \n' % + limit_len(get_properties(annotation))) + s.write('
{vtt_content}"
+
+
+class OCRTab(AnnotationTab):
+ def __init__(self, mmif, view, viz_id):
+ self.viz_id = viz_id
+ self.vid_path = mmif.get_documents_by_type(DocumentTypes.VideoDocument)[
+ 0].location_path()
+
+ super().__init__(mmif, view)
+
+ def render(self):
+ return render_template("pre-ocr.html", view_id=self.view.id, tabname=self.tab_name, mmif_id=self.viz_id)
+ # prepare_ocr(self.mmif, self.view, self.viz_id)
+ # return render_ocr_page(self.viz_id, self.vid_path, self.view.id, 0)
def render_ocr_page(mmif_id, vid_path, view_id, page_number):
"""
Renders a single OCR page by iterating through frames and displaying the
- contents/alignments.
+ contents/alignments. Note: this needs to be a separate function (not a method
+ in OCRTab) because it is called by the server when the page is changed.
"""
# Path for storing temporary images generated by cv2
cv2_vid = cv2.VideoCapture(vid_path)