def recurse_nodes_to_extract_knowledge_map(node, node_cache):
"""
Internal function for recursing the topic tree and building the knowledge map.
Requires rebranding of metadata done by recurse_nodes function.
"""
assert node["kind"] == "Topic"
if node.get("in_knowledge_map", None):
if node["slug"] not in knowledge_map["topics"]:
logging.debug("Not in knowledge map: %s" % node["slug"])
node["in_knowledge_map"] = False
for node in node_cache["Topic"][node["slug"]]:
node["in_knowledge_map"] = False
knowledge_topics[node["slug"]] = topic_tools.get_all_leaves(node, leaf_type="Exercise")
if not knowledge_topics[node["slug"]]:
sys.stderr.write("Removing topic from topic tree: no exercises. %s" % node["slug"])
del knowledge_topics[node["slug"]]
del knowledge_map["topics"][node["slug"]]
node["in_knowledge_map"] = False
for node in node_cache["Topic"][node["slug"]]:
node["in_knowledge_map"] = False
else:
if node["slug"] in knowledge_map["topics"]:
sys.stderr.write("Removing topic from topic tree; does not belong. '%s'" % node["slug"])
logging.warn("Removing from knowledge map: %s" % node["slug"])
del knowledge_map["topics"][node["slug"]]
for child in [n for n in node.get("children", []) if n["kind"] == "Topic"]:
recurse_nodes_to_extract_knowledge_map(child, node_cache)
def validate_data(topictree, node_cache, slug2id_map, knowledge_map):
# Validate related videos
for exercise_nodes in node_cache['Exercise'].values():
exercise = exercise_nodes[0]
exercise_path = os.path.join(settings.PROJECT_PATH, "static", "js", "khan-exercises", "exercises", "%s.html" % exercise["slug"])
if not os.path.exists(exercise_path):
sys.stderr.write("Could not find exercise HTML file: %s\n" % exercise_path)
for vid_slug in exercise.get("related_video_slugs", []):
if vid_slug not in slug2id_map or slug2id_map[vid_slug] not in node_cache["Video"]:
sys.stderr.write("Could not find related video %s in node_cache (from exercise %s)\n" % (vid_slug, exercise["slug"]))
# Validate related exercises
for video_nodes in node_cache["Video"].values():
video = video_nodes[0]
ex = video["related_exercise"]
if ex and ex["slug"] not in node_cache["Exercise"]:
sys.stderr.write("Could not find related exercise %s in node_cache (from video %s)\n" % (ex["slug"], video["slug"]))
# Validate all topics have leaves
for topic_nodes in node_cache["Topic"].values():
topic = topic_nodes[0]
if not topic_tools.get_topic_by_path(topic["path"], root_node=topictree).get("children"):
sys.stderr.write("Could not find any children for topic %s\n" % (topic["path"]))
# Validate all topics in knowledge map are in the node cache
for slug in knowledge_map["topics"]:
if slug not in node_cache["Topic"]:
sys.stderr.write("Unknown topic in knowledge map: %s\n" % slug)
topicdata_path = os.path.join(settings.PROJECT_PATH + "/static/data/", "topicdata", "%s.json" % slug)
if not os.path.exists(topicdata_path):
sys.stderr.write("Could not find topic data in topicdata directory: '%s'\n" % slug)
# Validate all topics in node-cache are in (or out) of knowledge map, as requested.
for topic_nodes in node_cache["Topic"].values():
topic = topic_nodes[0]
if topic["in_knowledge_map"] and not topic["slug"] in knowledge_map["topics"]:
sys.stderr.write("Topic '%-40s' not in knowledge map, but node_cache says it should be.\n" % topic["slug"])
elif not topic["in_knowledge_map"] and topic["slug"] in knowledge_map["topics"]:
sys.stderr.write("Topic '%-40s' in knowledge map, but node_cache says it shouldn't be.\n" % topic["slug"])
elif topic["in_knowledge_map"] and not topic_tools.get_topic_by_path(topic["path"], root_node=topictree).get("children"):
sys.stderr.write("Topic '%-40s' in knowledge map, but has no children.\n" % topic["slug"])
elif topic["in_knowledge_map"] and not topic_tools.get_all_leaves(topic_tools.get_topic_by_path(topic["path"], root_node=topictree), leaf_type="Exercise"):
sys.stderr.write("Topic '%40s' in knowledge map, but has no exercises.\n" % topic["slug"])
def recurse_nodes_to_extract_knowledge_map(node, node_cache):
"""
Internal function for recursing the topic tree and building the knowledge map.
Requires rebranding of metadata done by recurse_nodes function.
"""
assert node["kind"] == "Topic"
if node.get("in_knowledge_map", None):
if node["slug"] not in knowledge_map["topics"]:
logging.debug("Not in knowledge map: %s" % node["slug"])
node["in_knowledge_map"] = False
for node in node_cache["Topic"][node["slug"]]:
node["in_knowledge_map"] = False
knowledge_topics[node["slug"]] = topic_tools.get_all_leaves(
node, leaf_type="Exercise")
if not knowledge_topics[node["slug"]]:
sys.stderr.write(
"Removing topic from topic tree: no exercises. %s" %
node["slug"])
del knowledge_topics[node["slug"]]
del knowledge_map["topics"][node["slug"]]
node["in_knowledge_map"] = False
for node in node_cache["Topic"][node["slug"]]:
node["in_knowledge_map"] = False
else:
if node["slug"] in knowledge_map["topics"]:
sys.stderr.write(
"Removing topic from topic tree; does not belong. '%s'" %
node["slug"])
logging.warn("Removing from knowledge map: %s" % node["slug"])
del knowledge_map["topics"][node["slug"]]
for child in [
n for n in node.get("children", []) if n["kind"] == "Topic"
]:
recurse_nodes_to_extract_knowledge_map(child, node_cache)
def validate_data(topictree, node_cache, knowledge_map):
# Validate related videos
for exercise in node_cache['Exercise'].values():
exercise_path = os.path.join(settings.PROJECT_PATH, "static", "js",
"khan-exercises", "exercises",
"%s.html" % exercise["slug"])
if not os.path.exists(exercise_path):
sys.stderr.write("Could not find exercise HTML file: %s\n" %
exercise_path)
for vid in exercise.get("related_video_readable_ids", []):
if not vid in node_cache["Video"]:
sys.stderr.write(
"Could not find related video %s in node_cache (from exercise %s)\n"
% (vid, exercise["slug"]))
# Validate related exercises
for video in node_cache["Video"].values():
ex = video["related_exercise"]
if ex and not ex["slug"] in node_cache["Exercise"]:
sys.stderr.write(
"Could not find related exercise %s in node_cache (from video %s)\n"
% (ex["slug"], video["slug"]))
# Validate all topics have leaves
for topic in node_cache["Topic"].values():
if not topic_tools.get_topic_by_path(
topic["path"], root_node=topictree).get("children"):
sys.stderr.write("Could not find any children for topic %s\n" %
(topic["path"]))
# Validate all topics in knowledge map are in the node cache
for slug in knowledge_map["topics"]:
if slug not in node_cache["Topic"]:
sys.stderr.write("Unknown topic in knowledge map: %s\n" % slug)
topicdata_path = os.path.join(settings.PROJECT_PATH + "/static/data/",
"topicdata", "%s.json" % slug)
if not os.path.exists(topicdata_path):
sys.stderr.write(
"Could not find topic data in topicdata directory: '%s'\n" %
slug)
# Validate all topics in node-cache are in (or out) of knowledge map, as requested.
for topic in node_cache["Topic"].values():
if topic["in_knowledge_map"] and not topic["slug"] in knowledge_map[
"topics"]:
sys.stderr.write(
"Topic '%-40s' not in knowledge map, but node_cache says it should be.\n"
% topic["slug"])
elif not topic["in_knowledge_map"] and topic["slug"] in knowledge_map[
"topics"]:
sys.stderr.write(
"Topic '%-40s' in knowledge map, but node_cache says it shouldn't be.\n"
% topic["slug"])
elif topic["in_knowledge_map"] and not topic_tools.get_topic_by_path(
topic["path"], root_node=topictree).get("children"):
sys.stderr.write(
"Topic '%-40s' in knowledge map, but has no children.\n" %
topic["slug"])
elif topic["in_knowledge_map"] and not topic_tools.get_all_leaves(
topic_tools.get_topic_by_path(topic["path"],
root_node=topictree),
leaf_type="Exercise"):
sys.stderr.write(
"Topic '%40s' in knowledge map, but has no exercises.\n" %
topic["slug"])