Project History Blobs Collection

projectHistoryBlobs holds one document per project history: the metadata (hash and sizes) of every blob history-v1 has stored for that project. A blob is the content of a file or of a doc at some point of its history. The bytes themselves are objects in the project blobs bucket. There is no Mongoose schema: history-v1 writes the collection in services/history-v1/storage/lib/blob_store/mongo.js, which it uses for history ids of 24 hex characters, as in CE (services/history-v1/storage/lib/blob_store/index.js:108-116).

Fields

Field	Type	Meaning	Source
_id	ObjectId	History id of the project, the value in projects.overleaf.history.id.	mongo.js:37-40
blobs	Object	Blob records grouped in buckets. The key of a bucket is the first 3 hex digits of the hash.	mongo.js:39, 429-431
blobs.<abc>	Array of Object	Up to 8 records whose hash starts with abc. When a bucket already has 8, the next record goes to projectHistoryShardedBlobs.	mongo.js:23, 375-389
blobs.<abc>[].h	Binary	The hash, 20 bytes: git blob SHA-1, sha1("blob " + byteLength + "\0" + content) (libraries/overleaf-editor-core/lib/blob_utils.js:36-43). Its hex form is the hash used elsewhere, for example fileRefs.hash in projects.	mongo.js:472
blobs.<abc>[].b	Number	Byte length of the content.	mongo.js:473
blobs.<abc>[].s	Number	Length of the content as text; null when the content is not editable text (binary, or too large to edit) (blob_utils.js:116-130).	mongo.js:474

Who Writes It

Code	What it does
services/history-v1/storage/lib/blob_store/mongo.js:34-47	initialize: inserts { _id, blobs: {} } when history-v1 creates the project history (services/history-v1/storage/lib/chunk_store/index.js:67-68).
mongo.js:370-390	insertBlob: $addToSet of the record into its bucket while the bucket has fewer than 8 records, otherwise into the sharded collection. Runs after the bytes are uploaded, from putString, putBlob and copyBlob (blob_store/index.js:223-237, 271-278, 456-467).
mongo.js:54-105	clone: copies the buckets to another history id, for history-v1 POST /projects/:project_id/clone. Nothing in web calls that route (see Project History Chunks).
mongo.js:413-422	deleteBlobs: deletes the document and the project’s sharded documents, on history-v1 DELETE /projects/:project_id (services/history-v1/api/controllers/projects.js:361-373).

Where new blobs come from:

• File uploads: web sends the file to history-v1 PUT /projects/:project_id/blobs/:hash (services/web/app/src/Features/History/HistoryManager.mjs:164-169, services/history-v1/api/controllers/projects.js:376-402).
• Project copies: web asks history-v1 to copy each file’s blob with POST /projects/:project_id/blobs/:hash?copyFrom= (services/web/app/src/Features/Project/ProjectDuplicator.mjs:333-338, HistoryManager.mjs:177-187, projects.js:521-555).
• Doc text: when history-v1 writes a chunk, it stores the text of each doc, and its comments and tracked changes as JSON, as blobs (chunk_store/index.js:402-405, libraries/overleaf-editor-core/lib/file_data/string_file_data.js:166-173).

Who Reads It

Code	What it does
mongo.js:113-157	findBlob: one blob by hash; looks in the sharded collection when the bucket is full. Used by BlobStore.getBlob (blob_store/index.js:376-387) for the blob routes (projects.js:416-480), and before an insert to skip blobs that already exist (blob_store/index.js:441-448).
mongo.js:165-260	findBlobs: many blobs at once, for loading the files of a chunk.
mongo.js:265-309	getProjectBlobs: all blobs of a project.
mongo.js:316-363	getProjectBlobsBatch: all blobs of several projects, for history-v1 POST /projects/blob-stats (projects.js:620). The admin-tools module shows the result as the project size (services/web/modules/admin-tools/app/src/AdminProjectManager.mjs:395-405).

Indexes

Only the default _id index. No migration in tools/migrations/ adds one.

Notes

• The bytes of a blob are at the key <project key>/<hash[0:2]>/<hash[2:]> (blob_store/index.js:51-53) in the bucket OVERLEAF_EDITOR_PROJECT_BLOBS_BUCKET (services/history-v1/config/custom-environment-variables.json:49-52). The project key is the zero-padded history id, reversed and split as 3/3/rest characters (libraries/object-persistor/src/ProjectKey.js:8-15).
• A blob is stored once per project: an insert of a hash the project already has is skipped (blob_store/index.js:227-230, 249-252).
• The empty content needs no record; getBlob answers it without a lookup (blob_store/index.js:378-380).
• projectHistoryGlobalBlobs is read at start-up (blob_store/index.js:138-148) but no CE code writes it, so in CE every blob is a project blob.