reading-logseq-data

Reading Logseq Data

When to Use This Skill

This skill auto-invokes when:

• User wants to read pages or blocks from their Logseq graph
• Fetching properties or metadata from Logseq entities
• Executing Datalog queries against the graph
• Searching for content in Logseq
• Finding backlinks or references
• User mentions "get from logseq", "fetch page", "query logseq"

Client Library: See {baseDir}/scripts/logseq-client.py for the unified API.

Available Operations

Operation	Description
get_page(title)	Get page content and properties
get_block(uuid)	Get block with children
search(query)	Full-text search across graph
datalog_query(query)	Execute Datalog query
list_pages()	List all pages
get_backlinks(title)	Find pages linking to this one
get_graph_info()	Get current graph metadata

Quick Examples

Get a Page

from logseq_client import LogseqClient

client = LogseqClient()
page = client.get_page("My Page")
print(f"Title: {page['title']}")
print(f"Properties: {page['properties']}")

Execute Datalog Query

# Find all books with rating >= 4
results = client.datalog_query('''
    [:find (pull ?b [:block/title :user.property/rating])
     :where
     [?b :block/tags ?t]
     [?t :block/title "Book"]
     [?b :user.property/rating ?r]
     [(>= ?r 4)]]
''')

for book in results:
    print(f"{book['block/title']}: {book['user.property/rating']} stars")

Search Content

# Search for mentions of "project"
results = client.search("project")
for block in results:
    print(f"Found in: {block['page']}")
    print(f"Content: {block['content'][:100]}...")

Datalog Query Patterns

Find All Pages

[:find (pull ?p [:block/title])
 :where
 [?p :block/tags ?t]
 [?t :db/ident :logseq.class/Page]]

Find Blocks with Tag

[:find (pull ?b [*])
 :where
 [?b :block/tags ?t]
 [?t :block/title "Book"]]

Find by Property

[:find ?title ?author
 :where
 [?b :block/title ?title]
 [?b :user.property/author ?author]
 [?b :block/tags ?t]
 [?t :block/title "Book"]]

Find Tasks by Status

[:find (pull ?t [:block/title :logseq.property/status])
 :where
 [?t :block/tags ?tag]
 [?tag :db/ident :logseq.class/Task]
 [?t :logseq.property/status ?s]
 [?s :block/title "In Progress"]]

Find Backlinks

[:find (pull ?b [:block/title {:block/page [:block/title]}])
 :in $ ?page-title
 :where
 [?p :block/title ?page-title]
 [?b :block/refs ?p]]

Aggregations

;; Count books per author
[:find ?author (count ?b)
 :where
 [?b :block/tags ?t]
 [?t :block/title "Book"]
 [?b :user.property/author ?author]]

Using the Client Library

Initialization

from logseq_client import LogseqClient

# Auto-detect backend
client = LogseqClient()

# Force specific backend
client = LogseqClient(backend="http")

# Custom URL/token
client = LogseqClient(
    url="http://localhost:12315",
    token="your-token"
)

Error Handling

try:
    page = client.get_page("Nonexistent Page")
except client.NotFoundError:
    print("Page doesn't exist")
except client.ConnectionError:
    print("Cannot connect to Logseq")
except client.AuthError:
    print("Invalid token")

Batch Operations

# Get multiple pages efficiently
pages = ["Page1", "Page2", "Page3"]
results = [client.get_page(p) for p in pages]

# Or use a single query
query = '''
    [:find (pull ?p [*])
     :in $ [?titles ...]
     :where
     [?p :block/title ?titles]]
'''
results = client.datalog_query(query, [pages])

Performance Tips

1. Use specific queries - Don't fetch more than needed
2. Prefer pull syntax - (pull ?e [:needed :fields]) vs [*]
3. Put selective clauses first - Filter early in query
4. Use parameters - Pass values via :in clause
5. Batch when possible - Multiple items in one query

CLI Fallback

If HTTP API unavailable, the client falls back to CLI:

# CLI mode (automatic if HTTP fails)
client = LogseqClient(backend="cli", graph_path="/path/to/graph")

# Query still works the same way
results = client.datalog_query("[:find ?title :where [?p :block/title ?title]]")

Output Formats

Raw (default)

Returns Python dicts/lists directly from API.

Normalized

# Get normalized output
page = client.get_page("My Page", normalize=True)
# Returns: {"title": "...", "uuid": "...", "properties": {...}, "blocks": [...]}

Reference Materials

• See {baseDir}/references/read-operations.md for all operations
• See {baseDir}/templates/query-template.edn for query patterns