Repository navigation
Expand file tree
/
Copy pathmultimodal_test.py
More file actions
526 lines (449 loc) · 19 KB
/
Copy pathmultimodal_test.py
File metadata and controls
526 lines (449 loc) · 19 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
# Copyright 2025 Google LLC
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
def test_multimodal_dataframe(gcs_bucket_snippets: str) -> None:
# destination folder must be in a GCS bucket that the BQ connection service account (default or user provided) has write access to.
dst_bucket = f"gs://{gcs_bucket_snippets}"
# [START bigquery_dataframes_multimodal_dataframe_create]
import bigframes
# Flags to control preview image/video preview size
bigframes.options.display.blob_display_width = 300
import bigframes.pandas as bpd
# Create blob columns from wildcard path.
df_image = bpd._from_glob_path(
"gs://cloud-samples-data/bigquery/tutorials/cymbal-pets/images/*", name="image"
)
# Other ways are: from string uri column
# df = bpd.DataFrame({"uri": ["gs://<my_bucket>/<my_file_0>", "gs://<my_bucket>/<my_file_1>"]})
# df["blob_col"] = df["uri"].str.to_blob()
# From an existing object table
# df = bpd.read_gbq_object_table("<my_object_table>", name="blob_col")
# Take only the 5 images to deal with. Preview the content of the Mutimodal DataFrame
df_image = df_image.head(5)
df_image
# [END bigquery_dataframes_multimodal_dataframe_create]
# [START bigquery_dataframes_multimodal_dataframe_merge]
# Combine unstructured data with structured data
df_image["author"] = ["alice", "bob", "bob", "alice", "bob"] # type: ignore
df_image["content_type"] = df_image["image"].blob.content_type()
df_image["size"] = df_image["image"].blob.size()
df_image["updated"] = df_image["image"].blob.updated()
df_image
# [END bigquery_dataframes_multimodal_dataframe_merge]
# [START bigquery_dataframes_multimodal_dataframe_filter]
# Filter images and display, you can also display audio and video types. Use width/height parameters to constrain window sizes.
df_image[df_image["author"] == "alice"]["image"].blob.display()
# [END bigquery_dataframes_multimodal_dataframe_filter]
# [START bigquery_dataframes_multimodal_dataframe_image_transform]
df_image["blurred"] = df_image["image"].blob.image_blur(
(20, 20), dst=f"{dst_bucket}/image_blur_transformed/", engine="opencv"
)
df_image["resized"] = df_image["image"].blob.image_resize(
(300, 200), dst=f"{dst_bucket}/image_resize_transformed/", engine="opencv"
)
df_image["normalized"] = df_image["image"].blob.image_normalize(
alpha=50.0,
beta=150.0,
norm_type="minmax",
dst=f"{dst_bucket}/image_normalize_transformed/",
engine="opencv",
)
# You can also chain functions together
df_image["blur_resized"] = df_image["blurred"].blob.image_resize(
(300, 200), dst=f"{dst_bucket}/image_blur_resize_transformed/", engine="opencv"
)
df_image
# [END bigquery_dataframes_multimodal_dataframe_image_transform]
# [START bigquery_dataframes_multimodal_dataframe_ml_text]
from bigframes.ml import llm
gemini = llm.GeminiTextGenerator(model_name="gemini-2.0-flash-001")
# Deal with first 2 images as example
df_image = df_image.head(2)
# Ask the same question on the images
df_image = df_image.head(2)
answer = gemini.predict(df_image, prompt=["what item is it?", df_image["image"]])
answer[["ml_generate_text_llm_result", "image"]]
# [END bigquery_dataframes_multimodal_dataframe_ml_text]
# [START bigquery_dataframes_multimodal_dataframe_ml_text_alt]
# Ask different questions
df_image["question"] = [ # type: ignore
"what item is it?",
"what color is the picture?",
]
answer_alt = gemini.predict(
df_image, prompt=[df_image["question"], df_image["image"]]
)
answer_alt[["ml_generate_text_llm_result", "image"]]
# [END bigquery_dataframes_multimodal_dataframe_ml_text_alt]
# [START bigquery_dataframes_multimodal_dataframe_ml_embed]
# Generate embeddings on images
embed_model = llm.MultimodalEmbeddingGenerator()
embeddings = embed_model.predict(df_image["image"])
embeddings
# [END bigquery_dataframes_multimodal_dataframe_ml_embed]
# [START bigquery_dataframes_multimodal_dataframe_pdf_chunk]
# PDF chunking
df_pdf = bpd._from_glob_path(
"gs://cloud-samples-data/bigquery/tutorials/cymbal-pets/documents/*", name="pdf"
)
df_pdf["chunked"] = df_pdf["pdf"].blob.pdf_chunk(engine="pypdf")
chunked = df_pdf["chunked"].explode()
chunked
# [END bigquery_dataframes_multimodal_dataframe_pdf_chunk]
assert df_image is not None
assert answer is not None
assert answer_alt is not None
assert embeddings is not None
assert chunked is not None
def test_multimodal_example(gcs_bucket_snippets: str) -> None:
BUCKET = gcs_bucket_snippets
# [START bigquery_dataframes_multimodal_load]
import bigframes.bigquery as bbq
import bigframes.pandas as bpd
bbq.load_data(
"cymbal_pets.products",
write_disposition="OVERWRITE",
from_files_options={
"format": "avro",
"uris": [
"gs://cloud-samples-data/bigquery/tutorials/cymbal-pets/tables/products/products_*.avro"
],
},
)
# [END bigquery_dataframes_multimodal_load]
# [START bigquery_dataframes_multimodal_create_images]
bbq.create_external_table(
"cymbal_pets.product_images",
replace=True,
connection_name="us.cymbal_conn",
options={
"object_metadata": "SIMPLE",
"uris": [
"gs://cloud-samples-data/bigquery/tutorials/cymbal-pets/images/*.png"
],
},
)
# [END bigquery_dataframes_multimodal_create_images]
# [START bigquery_dataframes_multimodal_create_manuals]
bbq.create_external_table(
"cymbal_pets.product_manuals",
replace=True,
connection_name="us.cymbal_conn",
options={
"object_metadata": "SIMPLE",
"uris": [
"gs://cloud-samples-data/bigquery/tutorials/cymbal-pets/documents/*.pdf"
],
},
)
# [END bigquery_dataframes_multimodal_create_manuals]
# [START bigquery_dataframes_multimodal_create_gemini]
gemini_model = bbq.ml.create_model(
"cymbal_pets.gemini",
replace=True,
connection_name="us.cymbal_conn",
options={"endpoint": "gemini-3.5-flash"},
)
# [END bigquery_dataframes_multimodal_create_gemini]
# [START bigquery_dataframes_multimodal_create_embedding]
embedding_model = bbq.ml.create_model(
"cymbal_pets.embedding_model",
replace=True,
connection_name="us.cymbal_conn",
options={"endpoint": "gemini-embedding-2"},
)
# [END bigquery_dataframes_multimodal_create_embedding]
# [START bigquery_dataframes_multimodal_create_df_products_mm]
df_images = bpd.read_gbq("SELECT * FROM cymbal_pets.product_images")
df_products = bpd.read_gbq("cymbal_pets.products")
df_products_mm = df_images.merge(df_products, on="uri").drop(columns="uri")
df_products_mm = df_products_mm.rename(columns={"ref": "image"})
# [END bigquery_dataframes_multimodal_create_df_products_mm]
# [START bigquery_dataframes_multimodal_show_df_products_mm]
df_products_mm[["product_name", "image"]]
# [END bigquery_dataframes_multimodal_show_df_products_mm]
# [START bigquery_dataframes_multimodal_image_description]
df_products_mm["url"] = bbq.obj.get_access_url(
df_products_mm["image"], "R"
).to_frame()
df_products_mm["prompt0"] = "Can you describe the following image?"
df_products_mm["prompt"] = bbq.struct(df_products_mm[["prompt0", "url"]])
df_products_mm = bbq.ai.generate_table(
gemini_model, df_products_mm, output_schema={"image_description": "STRING"}
)
df_products_mm = df_products_mm[
[
"product_id",
"product_name",
"brand",
"category",
"subcategory",
"animal_type",
"search_keywords",
"price",
"description",
"inventory_level",
"supplier_id",
"average_rating",
"image",
"image_description",
]
]
# [END bigquery_dataframes_multimodal_image_description]
# [START bigquery_dataframes_multimodal_generate_animal_type]
df_prompt = bbq.obj.get_access_url(df_products_mm["image"], "R").to_frame()
df_prompt[
"prompt0"
] = "For the image of a pet product, concisely generate the following metadata: 1) animal_type and 2) 5 SEO search keywords, and 3) product subcategory."
df_products_mm["prompt"] = bbq.struct(df_prompt[["prompt0", "image"]])
df_products_mm = df_products_mm.drop(
columns=["animal_type", "search_keywords", "subcategory"]
)
df_products_mm = bbq.ai.generate_table(
gemini_model,
df_products_mm,
output_schema="animal_type STRING, search_keywords ARRAY<STRING>, subcategory STRING",
)
# [END bigquery_dataframes_multimodal_generate_animal_type]
# [START bigquery_dataframes_multimodal_show_animal_type]
df_products_mm[
[
"product_name",
"image_description",
"animal_type",
"search_keywords",
"subcategory",
]
]
# [END bigquery_dataframes_multimodal_show_animal_type]
# [START bigquery_dataframes_multimodal_brand_description]
df_agg = df_products_mm[
["image", "description", "category", "subcategory", "brand"]
]
df_agg["image"] = bbq.obj.get_access_url(df_products_mm["image"], "R")
df_agg = bbq.array_agg(df_agg.groupby(by=["brand"]))
df_agg["cnt"] = bbq.array_length(df_agg["image"])
df_prompt = df_agg[["image", "description", "category", "subcategory"]]
df_prompt[
"prompt0"
] = "Use the images and text to give one concise brand description for a website brand page. Return the description only. "
df_agg["prompt"] = bbq.struct(
df_prompt[["prompt0", "image", "description", "category", "subcategory"]]
)
df_agg = df_agg.reset_index()
df_agg = bbq.ai.generate_table(
gemini_model, df_agg, output_schema={"brand_description": "STRING"}
)
df_agg[["brand", "brand_description", "cnt"]]
# [END bigquery_dataframes_multimodal_brand_description]
# [START bigquery_dataframes_multimodal_define_to_grayscale]
@bpd.udf(
dataset="cymbal_pets",
name="to_grayscale",
packages=["numpy", "opencv-python"],
bigquery_connection="us.cymbal_conn",
max_batching_rows=1,
)
def to_grayscale(src_ref: str, dst_ref: str) -> str:
import json
from urllib.request import Request, urlopen
import cv2 as cv
import numpy as np
src_json = json.loads(src_ref)
srcUrl = src_json["access_urls"]["read_url"]
dst_json = json.loads(dst_ref)
dstUrl = dst_json["access_urls"]["write_url"]
req = urlopen(srcUrl)
arr = np.asarray(bytearray(req.read()), dtype=np.uint8)
img = cv.imdecode(arr, -1) # 'Load it as it is'
# Convert the image to grayscale
gray_image = cv.cvtColor(img, cv.COLOR_BGR2GRAY)
# Send POST request to the URL
_, img_encoded = cv.imencode(".png", gray_image)
req = Request(
url=dstUrl,
data=img_encoded.tobytes(),
method="PUT",
headers={
"Content-Type": "image/png",
},
)
with urlopen(req):
pass
return dst_ref
# [END bigquery_dataframes_multimodal_define_to_grayscale]
# [START bigquery_dataframes_multimodal_apply_to_grayscale]
df_grayscale = df_products_mm[["product_id", "product_name", "image"]]
df_grayscale[
"gray_image_uri"
] = f"gs://{BUCKET}/cymbal-pets-images/grayscale/" + df_grayscale[
"image"
].struct.field(
"uri"
).str.extract(
r"([^/]+)$"
)
df_grayscale["gray_image"] = bbq.obj.make_ref(
df_grayscale["gray_image_uri"], "us.cymbal_conn"
)
df_grayscale["image_url"] = bbq.to_json_string(
bbq.obj.get_access_url(df_grayscale["image"], "r")
)
df_grayscale["gray_image_url"] = bbq.to_json_string(
bbq.obj.get_access_url(df_grayscale["gray_image"], "rw")
)
df_grayscale[["image_url", "gray_image_url"]].apply(to_grayscale, axis=1)
# [END bigquery_dataframes_multimodal_apply_to_grayscale]
# [START bigquery_dataframes_multimodal_define_chunk_pdf]
@bpd.udf(
dataset="cymbal_pets",
name="chunk_pdf",
packages=["pypdf"],
bigquery_connection="us.cymbal_conn",
max_batching_rows=1,
)
def chunk_pdf(src_ref: str, chunk_size: int, overlap_size: int) -> list[str]:
import io
import json
from urllib.request import urlopen
from pypdf import PdfReader # type: ignore
src_json = json.loads(src_ref)
srcUrl = src_json["access_urls"]["read_url"]
req = urlopen(srcUrl)
pdf_file = io.BytesIO(bytearray(req.read()))
reader = PdfReader(pdf_file, strict=False)
# extract and chunk text simultaneously
all_text_chunks = []
curr_chunk = ""
for page in reader.pages:
page_text = page.extract_text()
if page_text:
curr_chunk += page_text
# split the accumulated text into chunks of a specific size with overlaop
# this loop implements a sliding window approach to create chunks
while len(curr_chunk) >= chunk_size:
split_idx = curr_chunk.rfind(" ", 0, chunk_size)
if split_idx == -1:
split_idx = chunk_size
actual_chunk = curr_chunk[:split_idx]
all_text_chunks.append(actual_chunk)
overlap = curr_chunk[split_idx + 1 : split_idx + 1 + overlap_size]
curr_chunk = overlap + curr_chunk[split_idx + 1 + overlap_size :]
if curr_chunk:
all_text_chunks.append(curr_chunk)
return all_text_chunks
# [END bigquery_dataframes_multimodal_define_chunk_pdf]
# [START bigquery_dataframes_multimodal_apply_chunk_pdf]
df_manuals = bpd.read_gbq("SELECT * FROM cymbal_pets.product_manuals")
df_manuals["url"] = bbq.to_json_string(
bbq.obj.get_access_url(df_manuals["ref"], "R")
)
df_manuals["chunk_size"] = 1000
df_manuals["overlap_size"] = 100
df_manuals["chunked"] = df_manuals[["url", "chunk_size", "overlap_size"]].apply(
chunk_pdf, axis=1
)
# [END bigquery_dataframes_multimodal_apply_chunk_pdf]
# [START bigquery_dataframes_multimodal_analyze_pdf]
df_chunked = df_manuals["chunked"].explode().to_frame()
df_chunked[
"prompt0"
] = "Can you summarize the product manual as bullet points? Highlight the legal clauses"
df_chunked["prompt"] = bbq.struct(df_chunked[["prompt0", "chunked"]])
result = bbq.ai.generate_text(gemini_model, df_chunked["prompt"])
result
# [END bigquery_dataframes_multimodal_analyze_pdf]
# [START bigquery_dataframes_multimodal_create_embed_table]
df_products_mm["content"] = bbq.obj.get_access_url(df_products_mm["image"], "R")
df_embed = bbq.ai.generate_embedding(
embedding_model, df_products_mm[["content", "product_id"]]
)
df_embed.to_gbq("cymbal_pets.products_embedding", if_exists="replace")
# [END bigquery_dataframes_multimodal_create_embed_table]
# [START bigquery_dataframes_multimodal_vector_search]
df_image = bpd.DataFrame(
{
"uri": [
"gs://cloud-samples-data/bigquery/tutorials/cymbal-pets/images/cozy-naps-cat-scratching-post-with-condo.png"
]
}
).cache()
df_image["image"] = bbq.obj.make_ref(df_image["uri"], "us.cymbal_conn")
df_search = bbq.ai.generate_embedding(
embedding_model,
bbq.obj.get_access_url(bbq.obj.fetch_metadata(df_image["image"]), "R"),
)
search_result = bbq.vector_search(
"cymbal_pets.products_embedding", "embedding", df_search["embedding"]
)
search_result
# [END bigquery_dataframes_multimodal_vector_search]
# [START bigquery_dataframes_create_external_table_all]
bbq.create_external_table(
"cymbal_pets.product_manuals_all",
replace=True,
connection_name="us.cymbal_conn",
options={
"object_metadata": "SIMPLE",
"uris": [
"gs://cloud-samples-data/bigquery/tutorials/cymbal-pets/documents/*.pdf",
"gs://cloud-samples-data/bigquery/tutorials/cymbal-pets/document_chunks/*.pdf",
],
},
)
# [END bigquery_dataframes_create_external_table_all]
# [START bigquery_dataframes_create_manual_to_chunks]
df1 = bpd.read_gbq("SELECT * FROM cymbal_pets.product_manuals_all").sort_values(
"uri"
)
df2 = df1.copy()
df1["name"] = df1["uri"].str.extract(r".*/([^.]*).[^/]+")
df2["name"] = df2["uri"].str.extract(r".*/([^.]*)_page[0-9]+.[^/]+")
df_manuals_all = df1.merge(df2, on="name")
df_manuals_agg = (
bbq.array_agg(df_manuals_all[["ref_x", "uri_x"]].groupby("uri_x"))["ref_x"]
.str[0]
.to_frame()
)
df_manuals_agg["chunks"] = bbq.array_agg(
df_manuals_all[["ref_y", "uri_x"]].groupby("uri_x")
)["ref_y"]
# [END bigquery_dataframes_create_manual_to_chunks]
# [START bigquery_dataframes_show_manual_to_chunks]
df_manuals_agg
# [END bigquery_dataframes_show_manual_to_chunks]
# [START bigquery_dataframes_generate_pages_summary]
df_manuals_agg["chunks_url"] = bbq.array_agg(
bbq.obj.get_access_url(df_manuals_agg.explode("chunks")["chunks"], "R").groupby(
"uri_x"
)
)
df_manuals_agg[
"prompt0"
] = "Can you provide a page by page summary for the first 3 pages of the attached manual? Only write one line for each page. The pages are provided in serial order"
df_manuals_agg["prompt"] = bbq.struct(df_manuals_agg[["prompt0", "chunks_url"]])
result = bbq.ai.generate_text(gemini_model, df_manuals_agg["prompt"])["result"]
result
# [END bigquery_dataframes_generate_pages_summary]
# [START bigquery_dataframes_generate_each_page_summary]
result = bbq.ai.generate_table(
gemini_model,
df_manuals_agg["prompt"],
output_schema={
"page1_summary": "STRING",
"page2_summary": "STRING",
"page3_summary": "STRING",
},
)[["page1_summary", "page2_summary", "page3_summary"]]
result
# [END bigquery_dataframes_generate_each_page_summary]