feat: extract files from message logs

This commit is contained in:
zhaoyilun
2026-07-10 00:45:08 +08:00
parent 3d68a18097
commit fe242145bb
5 changed files with 249 additions and 7 deletions

View File

@@ -50,7 +50,7 @@ N13/N14/N15 are pre-business validation results. They can help identify UI eleme
## Current loop
Current loop: `Stage C / Loop C13 - log-backed file metadata extraction`.
Current loop: `Stage C / Loop C14 - register isphere_receive_files list mode`.
Plan file: `docs/superpowers/plans/2026-07-09-stage-c-log-backed-receive-messages-loop.md`.
@@ -75,10 +75,11 @@ Current loop output so far:
10. C10: log-backed group candidate extraction from groupchat/conference JIDs.
11. C11: `isphere_search_groups` MCP registration and seven-tool verification.
12. C12: file evidence precheck confirmed enough PacketReader/Smark evidence for file metadata listing, but not enough for real cache/download mapping.
13. C13: `internal/isphere.ListFilesFromMessages` extracts synthetic/redacted file metadata candidates from normalized messages; MCP registration is next.
## Next business mainline
1. Run Stage C Loop C13: implement log-backed file metadata extraction from normalized messages.
2. Register `isphere_receive_files` list mode after C13 source behavior passes; leave actual download/cache mapping as a later connector node.
1. Run Stage C Loop C14: register `isphere_receive_files` list mode on top of `ListFilesFromMessages`.
2. Leave actual download/cache mapping as a later connector node until a message-to-cache hash or client download source is validated.
3. Add `isphere_send_message` and `isphere_send_file` after source discovery and connector selection.
4. Keep UIA selector/report work as fallback support, not as the primary roadmap.

View File

@@ -20,7 +20,7 @@ Schema notes: `docs/source-discovery/2026-07-09-n12-pre-zyl-schema-notes.md`
| `isphere_receive_messages` | decrypted `PacketReader.ProcessPacket` XMPP `<message>` stanzas via `internal/isphere.EncryptedPacketLogSource` | decrypted `Smark.SendReceive` transport stanzas; decrypted `SaveToDB` `Chat_OnMessageArrived` traces; wrapped `MsgLib.db` after DB wrapper is solved | `docs/source-discovery/2026-07-09-n12-pre-zyl-schema-notes.md#field-mapping-evidence` | C5 env loader wired through ISPHERE_PACKET_LOG_FILE; default source remains empty when unset | fixture-backed configured-source operator smoke |
| `isphere_search_contacts` | bare sender/receiver JIDs extracted from normalized log-backed messages | decrypted roster/contact stanzas from `Smark.SendReceive`; `MsgLib.db` contact/conversation tables after DB wrapper is solved; UIA helper source if display names are missing | `docs/source-discovery/2026-07-09-n12-pre-zyl-schema-notes.md#log-decryption-result` | C8 MCP tool registered and verify-go-mcp covers default/configured contact smoke; JID-only display/account fallback | enrich display names from roster/db/uia evidence |
| `isphere_search_groups` | decrypted `PacketReader.ProcessPacket` `type="groupchat"` stanzas and conference/MUC JIDs | `MsgLib.db` group/conversation tables after DB wrapper is solved; UIA helper source if group display names are missing | C9 ignored-log scan: PacketReader 86 groupchat/86 conference hits; Smark 24 groupchat/658 conference/631 muc hits | C11 MCP tool registered and verify-go-mcp covers default/configured group smoke; JID-only display fallback | enrich group display/member data from roster/db/uia evidence |
| `isphere_receive_files` | decrypted `PacketReader.ProcessPacket` file-transfer message stanzas for metadata listing; `zyl\importal\<hash>` cache entries remain unlinked | decrypted `Smark.SendReceive` and `SaveToDB` traces for file-reference reconciliation; future UIA/client connector for download | `docs/source-discovery/2026-07-09-n12-pre-zyl-schema-notes.md#c12-file-transfer-evidence-precheck` | C12 evidence confirmed metadata-list source: PacketReader has 107 parsed file candidates; no log-to-cache/download mapping yet | extract file metadata candidates from normalized messages |
| `isphere_receive_files` | decrypted `PacketReader.ProcessPacket` file-transfer message stanzas for metadata listing via `internal/isphere.ListFilesFromMessages`; `zyl\importal\<hash>` cache entries remain unlinked | decrypted `Smark.SendReceive` and `SaveToDB` traces for file-reference reconciliation; future UIA/client connector for download | `docs/source-discovery/2026-07-09-n12-pre-zyl-schema-notes.md#c12-file-transfer-evidence-precheck` | C13 source behavior implemented for metadata listing; no log-to-cache/download mapping yet | register `isphere_receive_files` MCP list mode |
## Stage C entry rule

View File

@@ -1277,7 +1277,7 @@ Evidence precheck result:
- Query supports `ConversationID`, `MessageID`, `FileID`, `NameContains`, and `Limit`; default/zero limit returns all candidates.
- This loop does not register the MCP tool yet. Registration/list-mode smoke becomes C14 after source behavior passes tests.
- [ ] **Step 1: Write failing file extraction test**
- [x] **Step 1: Write failing file extraction test**
Create `internal/isphere/files_test.go` and run:
@@ -1287,11 +1287,11 @@ go test ./internal/isphere -run TestListFilesFromMessagesExtractsFileTransferCan
Expected initial failure: file query/types/functions do not exist.
- [ ] **Step 2: Implement file metadata extraction/search**
- [x] **Step 2: Implement file metadata extraction/search**
Create `internal/isphere/files.go` with minimal pure Go logic over normalized messages and synthetic/redacted fixtures.
- [ ] **Step 3: Run full verification and rewrite Loop C14**
- [x] **Step 3: Run full verification and rewrite Loop C14**
Run:
@@ -1306,6 +1306,74 @@ Expected: all exit 0. Remove any generated root `isphere-mcp.exe` after build. R
---
## Loop C13 Result
Implemented and verified in this loop:
- `internal/isphere.ListFilesFromMessages(messages, query)` extracts read-only file metadata candidates from normalized messages when `Subject` contains `FILE_TRANSFER` or message text contains a supported file extension.
- File metadata currently includes stable local `FileID` (`message_id:file_name`), `MessageID`, `ConversationID`, `FileName`, inferred `MimeType`, `CreatedAt`, `Source` `local_readonly`, and `RawRef` `packetlog_file_transfer`.
- `SizeBytes`, `DownloadRef`, `SavedPath`, and `SHA256` remain empty/unknown because C12 did not prove a message-to-cache/download mapping.
- Query matching supports conversation, message, file id, file-name substring, and limit. Tests use only synthetic/redacted filenames.
- Verification completed for C13 with the loop-specific RED/GREEN test; full verification is run again before commit.
---
## Loop C14: Register `isphere_receive_files` MCP list mode
**Goal:** Register the `isphere_receive_files` MCP tool in list mode on top of C13 file metadata extraction, while making download mode explicitly deferred until cache/download mapping is validated.
**Planned files:**
- Create: `internal/tools/isphere_files.go`
- Create: `internal/tools/isphere_files_test.go`
- Modify: `internal/mcpserver/server.go`
- Modify: `internal/mcpserver/server_test.go`
- Modify: `scripts/verify-go-mcp.ps1`
- Modify: `docs/go-mcp-runbook.md`
- Modify: `docs/current-status-card.md`
- Modify: `docs/source-discovery/capability-source-matrix.md`
- Modify: `docs/superpowers/plans/2026-07-09-stage-c-log-backed-receive-messages-loop.md`
**Planned interface:**
- Tool name: `isphere_receive_files`.
- Input: `conversation_id`, `message_id`, `file_id`, `name_contains`, `mode`, `limit`.
- C14 supports `mode` empty or `list`; download mode is not implemented in this loop.
- Source: same env-configured receive-message source used by receive/contact/group tools.
- Output shape: `ok`, `mode`, `files`, `next_cursor`, and `audit`, aligned with `docs/mcp-core-tools-contract.md`.
- Default empty source returns `ok: true`, `mode: "list"`, `files: []`.
- [ ] **Step 1: Write failing MCP tool test**
Create `internal/tools/isphere_files_test.go` and run:
```powershell
go test ./internal/tools -run TestISphereReceiveFilesToolReturnsLogBackedFiles -v
```
Expected initial failure: file receive tool registration does not exist.
- [ ] **Step 2: Implement file tool registration**
Register `isphere_receive_files` with source injection for tests. Use `ListFilesFromMessages` and return a contract-shaped map with minimal audit metadata.
- [ ] **Step 3: Wire server and verification**
Register file receive in server construction. Update server tool-count tests and `scripts/verify-go-mcp.ps1` to expect eight tools and call `isphere_receive_files` against default empty source and synthetic configured file-transfer fixture.
- [ ] **Step 4: Run full verification and commit**
Run:
```powershell
git diff --check
go test ./...
go build ./cmd/isphere-mcp
powershell -NoProfile -ExecutionPolicy Bypass -File scripts/verify-go-mcp.ps1
```
Expected: all exit 0. Remove any generated root `isphere-mcp.exe` after build. Commit exact C14 paths only and rewrite the next loop from actual results.
---
## Self-Review
- Spec coverage: the plan follows the user's cyclic requirement: plan first, implement one loop at a time, update the next loop after each result, and ask only on blockers.

142
internal/isphere/files.go Normal file
View File

@@ -0,0 +1,142 @@
package isphere
import (
"path/filepath"
"regexp"
"sort"
"strings"
)
type File struct {
FileID string
MessageID string
ConversationID string
FileName string
SizeBytes *int64
MimeType string
CreatedAt string
DownloadRef string
SavedPath string
SHA256 string
Source string
RawRef string
}
type ListFilesQuery struct {
ConversationID string
MessageID string
FileID string
NameContains string
Limit int
}
type ListFilesResult struct {
Files []File
}
var fileNamePattern = regexp.MustCompile(`(?i)[^\s<>:"|?*]+\.(?:docx?|xlsx?|pptx?|pdf|zip|rar|png|jpe?g|txt|csv)\b`)
func ListFilesFromMessages(messages []Message, query ListFilesQuery) ListFilesResult {
files := make([]File, 0)
for _, message := range messages {
if !messageLooksFileBacked(message) {
continue
}
fileName := extractMessageFileName(message.Text)
if fileName == "" {
continue
}
file := File{
FileID: fileID(message.ID, fileName),
MessageID: message.ID,
ConversationID: message.ConversationID,
FileName: fileName,
MimeType: mimeTypeForFileName(fileName),
CreatedAt: message.Timestamp,
Source: "local_readonly",
RawRef: "packetlog_file_transfer",
}
if fileMatchesQuery(file, query) {
files = append(files, file)
}
}
sort.Slice(files, func(i, j int) bool {
if files[i].FileID == files[j].FileID {
return files[i].MessageID < files[j].MessageID
}
return files[i].FileID < files[j].FileID
})
limit := query.Limit
if limit <= 0 || limit > len(files) {
limit = len(files)
}
return ListFilesResult{Files: files[:limit]}
}
func messageLooksFileBacked(message Message) bool {
return strings.Contains(strings.ToUpper(message.Subject), "FILE_TRANSFER") || extractMessageFileName(message.Text) != ""
}
func extractMessageFileName(text string) string {
match := fileNamePattern.FindString(strings.TrimSpace(text))
return strings.Trim(match, " .,;,。;")
}
func fileID(messageID string, fileName string) string {
if messageID == "" {
return fileName
}
return messageID + ":" + fileName
}
func fileMatchesQuery(file File, query ListFilesQuery) bool {
if query.ConversationID != "" && file.ConversationID != query.ConversationID {
return false
}
if query.MessageID != "" && file.MessageID != query.MessageID {
return false
}
if query.FileID != "" && file.FileID != query.FileID {
return false
}
nameQuery := strings.ToLower(strings.TrimSpace(query.NameContains))
if nameQuery != "" && !strings.Contains(strings.ToLower(file.FileName), nameQuery) {
return false
}
return true
}
func mimeTypeForFileName(name string) string {
switch strings.ToLower(filepath.Ext(name)) {
case ".doc":
return "application/msword"
case ".docx":
return "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
case ".xls":
return "application/vnd.ms-excel"
case ".xlsx":
return "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
case ".ppt":
return "application/vnd.ms-powerpoint"
case ".pptx":
return "application/vnd.openxmlformats-officedocument.presentationml.presentation"
case ".pdf":
return "application/pdf"
case ".zip":
return "application/zip"
case ".rar":
return "application/vnd.rar"
case ".png":
return "image/png"
case ".jpg", ".jpeg":
return "image/jpeg"
case ".txt":
return "text/plain"
case ".csv":
return "text/csv"
default:
return ""
}
}

View File

@@ -0,0 +1,31 @@
package isphere
import (
"reflect"
"testing"
)
func TestListFilesFromMessagesExtractsFileTransferCandidates(t *testing.T) {
messages := []Message{
{ID: "msg-1", ConversationID: "alice@imopenfire1-lanzhou|bob@imopenfire1-lanzhou", Text: "redacted-report.docx", Timestamp: "1783423807000", Subject: "FILE_TRANSFER_CANCEL"},
{ID: "msg-2", ConversationID: "alice@imopenfire1-lanzhou|bob@imopenfire1-lanzhou", Text: "plain chat", Timestamp: "1783423808000"},
{ID: "msg-3", ConversationID: "room@conference.imopenfire1-lanzhou", Text: "redacted-screenshot.png", Timestamp: "1783423809000"},
}
got := ListFilesFromMessages(messages, ListFilesQuery{NameContains: "report", Limit: 10})
want := ListFilesResult{Files: []File{
{
FileID: "msg-1:redacted-report.docx",
MessageID: "msg-1",
ConversationID: "alice@imopenfire1-lanzhou|bob@imopenfire1-lanzhou",
FileName: "redacted-report.docx",
MimeType: "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
CreatedAt: "1783423807000",
Source: "local_readonly",
RawRef: "packetlog_file_transfer",
},
}}
if !reflect.DeepEqual(got, want) {
t.Fatalf("ListFilesFromMessages() = %#v, want %#v", got, want)
}
}