3 files changed, 52 insertions, 51 deletions
diff --git a/check/commands/phash/dedupe.py b/check/commands/phash/dedupe.py
index 3cf60d4..28266f4 100644
--- a/check/commands/phash/dedupe.py
+++ b/check/commands/phash/dedupe.py
@@ -1,5 +1,5 @@
 """
-Dedupe a folder of images
+Dedupe a folder of images (uses phash directly, does not use database)
 """
 
 import click
@@ -9,18 +9,22 @@ import glob
 from PIL import Image
 
 from app.utils.im_utils import compute_phash
+from app.utils.file_utils import write_json, sha256
 
 @click.command()
 @click.option('-i', '--input', 'opt_input_glob',
   required=True,
   help="Input glob to search -- e.g. '../docs/images/*.jpg'")
+@click.option('-o', '--output', 'opt_output_fn',
+  required=False,
+  help="Input glob to search -- e.g. '../docs/images/*.jpg'")
 @click.option('-t', '--threshold', 'opt_threshold',
   required=True,
   default=6,
   type=int,
   help="Threshold for PHash hamming distance comparison (0-64, default=6)")
 @click.pass_context
-def cli(ctx, opt_input_glob, opt_threshold):
+def cli(ctx, opt_input_glob, opt_output_fn, opt_threshold):
   """
   Dedupe a folder of images
   """
@@ -31,10 +35,14 @@ def cli(ctx, opt_input_glob, opt_threshold):
     im = Image.open(fn).convert('RGB')
     phash = compute_phash(im)
     if is_phash_new(fn, phash, seen, opt_threshold):
+      hash = sha256(fn)
       seen.append({
+        'sha256': hash,
         'phash': phash,
         'fn': fn,
       })
+  if opt_output_fn:
+    write_json(seen, opt_output_fn)
   print("checked {} files, found {} unique".format(total, len(seen)))
 
 def is_phash_new(fn, phash, seen, opt_threshold):
diff --git a/check/commands/phash/query.py b/check/commands/phash/query.py
index 8fc8c61..7fe2ae3 100644
--- a/check/commands/phash/query.py
+++ b/check/commands/phash/query.py
@@ -1,9 +1,10 @@
 """
-Search the database for an image
+Query the database with a test set
 """
 
 import click
 import os
+import glob
 
 from PIL import Image
 
@@ -12,34 +13,29 @@ from app.utils.im_utils import compute_phash_int
 from app.utils.file_utils import sha256
 
 @click.command()
-@click.option('-i', '--input', 'opt_fn',
+@click.option('-i', '--input', 'opt_input_glob',
   required=True,
-  help="File to search")
+  help="Input glob to search -- e.g. '../docs/images/*.jpg'")
 @click.pass_context
-def cli(ctx, opt_fn):
+def cli(ctx, opt_input_glob):
   """
-  Search the database for an image
+  Query the database with a test set
   """
-  print('Searching for a file...')
+  for fn in sorted(glob.iglob(opt_input_glob)):
+    im = Image.open(fn).convert('RGB')
+    phash = compute_phash_int(im)
 
-  if not os.path.exists(opt_fn):
-    print("File does not exist")
-    return
+    hash = sha256(fn)
 
-  im = Image.open(opt_fn).convert('RGB')
-  phash = compute_phash_int(im)
+    phash_match = search_by_phash(phash)
+    hash_match = search_by_hash(hash)
 
-  hash = sha256(opt_fn)
+    hash_result = 'NO'
+    if hash_match:
+      hash_result = 'YES'
 
-  phash_match = search_by_phash(phash)
-  hash_match = search_by_hash(hash)
+    phash_result = 'NO'
+    if len(phash_match):
+      phash_result = 'YES, score={}'.format(phash_match[0]['score'])
 
-  hash_result = 'NO'
-  if hash_match:
-    hash_result = 'YES'
-
-  phash_result = 'NO'
-  if len(phash_match):
-    phash_result = 'YES, score={}'.format(phash_match[0]['score'])
-
-  print("{} - hash={}, phash={}".format(opt_fn, hash_result, phash_result))
+    print("{} - hash={}, phash={}".format(fn, hash_result, phash_result))
diff --git a/check/commands/phash/test.py b/check/commands/phash/test.py
index 7fe2ae3..77c4c69 100644
--- a/check/commands/phash/test.py
+++ b/check/commands/phash/test.py
@@ -1,41 +1,38 @@
 """
-Query the database with a test set
+Test the API
 """
 
 import click
 import os
 import glob
+import requests
 
-from PIL import Image
-
-from app.models.sql_factory import search_by_phash, search_by_hash
-from app.utils.im_utils import compute_phash_int
-from app.utils.file_utils import sha256
+mime_types = {
+  '.png': 'image/png',
+  '.gif': 'image/gif',
+  '.jpg': 'image/jpeg',
+  '.jpeg': 'image/jpeg',
+}
 
 @click.command()
-@click.option('-i', '--input', 'opt_input_glob',
+@click.option('-i', '--input', 'opt_input_fn',
   required=True,
-  help="Input glob to search -- e.g. '../docs/images/*.jpg'")
+  help="Image to test the API with")
 @click.pass_context
-def cli(ctx, opt_input_glob):
+def cli(ctx, opt_input_fn):
   """
-  Query the database with a test set
+  Query the API with a test image
   """
-  for fn in sorted(glob.iglob(opt_input_glob)):
-    im = Image.open(fn).convert('RGB')
-    phash = compute_phash_int(im)
-
-    hash = sha256(fn)
-
-    phash_match = search_by_phash(phash)
-    hash_match = search_by_hash(hash)
-
-    hash_result = 'NO'
-    if hash_match:
-      hash_result = 'YES'
+  with open(opt_input_fn, 'rb') as f:
+    fn = os.path.basename(opt_input_fn)
+    fpart, ext = os.path.splitext(fn)
+    if ext not in mime_types:
+      print("Invalid filetype: {}".format(ext))
 
-    phash_result = 'NO'
-    if len(phash_match):
-      phash_result = 'YES, score={}'.format(phash_match[0]['score'])
+    query = [
+      ('q', (fn, f, mime_types[ext]))
+    ]
 
-    print("{} - hash={}, phash={}".format(fn, hash_result, phash_result))
+    print("Testing match API")
+    r = requests.post('http://0.0.0.0:5000/api/v1/match', files=query)
+    print(r.json())