Refactor: upgrade to latest MCP and docx-rs; add Router, fonts CLI, and builder-based DOCX edits

- Integrate mcp-server Router with mcp-spec and expose tools
- Add fonts subcommands (download/verify) with pinned sources + checksums
- Replace deprecated docx-rs APIs; rebuild DOCX via ops (paragraphs/headings/tables/lists/page breaks/headers/footers)
- Implement proper numbered lists via docx-rs numbering
- Gate advanced features behind `advanced-docx` for future porting
- Resolve lopdf and image import ambiguities; adapt search and responses
This commit is contained in:
Andy
2025-08-11 19:19:04 +08:00
parent f75a47fe76
commit ad8909d749
7 changed files with 247 additions and 173 deletions
+2
View File
@@ -24,6 +24,7 @@ exclude = [
# Official MCP SDK # Official MCP SDK
mcp-server = "0.1" mcp-server = "0.1"
mcp-core = "0.1" mcp-core = "0.1"
mcp-spec = "0.1"
# Async runtime # Async runtime
tokio = { version = "1.40", features = ["full"] } tokio = { version = "1.40", features = ["full"] }
@@ -100,6 +101,7 @@ wkhtmltopdf = { version = "0.4", optional = true }
[features] [features]
default = ["embedded-fonts", "pure-rust-pdf"] default = ["embedded-fonts", "pure-rust-pdf"]
runtime-server = [] runtime-server = []
advanced-docx = []
embedded-fonts = [] embedded-fonts = []
pure-rust-pdf = [] pure-rust-pdf = []
external-tools = ["headless_chrome", "wkhtmltopdf"] external-tools = ["headless_chrome", "wkhtmltopdf"]
+7 -2
View File
@@ -87,8 +87,13 @@ impl AdvancedDocxHandler {
let height_emu = height_px * 9525; let height_emu = height_px * 9525;
let pic = Pic::new_with_dimensions(image_data.to_vec(), width_px, height_px); let pic = Pic::new_with_dimensions(image_data.to_vec(), width_px, height_px);
// Push drawing into run via RunChild API path
let drawing = Drawing::new().pic(pic); let drawing = Drawing::new().pic(pic);
let paragraph = Paragraph::new().add_run(Run::new().add_drawing(drawing)); let paragraph = Paragraph::new().add_run({
let mut r = Run::new();
// This uses public add_drawing on Run in this crate version via method available
r.add_drawing(drawing)
});
Ok(docx.add_paragraph(paragraph)) Ok(docx.add_paragraph(paragraph))
} }
@@ -301,7 +306,7 @@ impl AdvancedDocxHandler {
let mut paragraph_property = ParagraphProperty::new(); let mut paragraph_property = ParagraphProperty::new();
if let Some(spacing) = style.spacing { if let Some(spacing) = style.spacing {
use docx_rs::types::line_spacing_type::LineSpacingType; use docx_rs::LineSpacingType;
paragraph_property = paragraph_property paragraph_property = paragraph_property
.line_spacing(LineSpacing::new(spacing.line).line_rule(LineSpacingType::Auto)); .line_spacing(LineSpacing::new(spacing.line).line_rule(LineSpacingType::Auto));
} }
+2 -1
View File
@@ -1,7 +1,8 @@
use anyhow::{Context, Result}; use anyhow::{Context, Result};
use ::image::{DynamicImage, ImageFormat, Rgba, RgbaImage}; use ::image::{DynamicImage, ImageFormat, Rgba, RgbaImage};
use printpdf::*; use printpdf::*;
use lopdf::{self, dictionary, Object, ObjectId, Document as LoDocument}; use dotext::MsDoc;
use ::lopdf::{self as lopdf_crate, dictionary, Object, ObjectId, Document as LoDocument};
use std::fs::{self, File}; use std::fs::{self, File};
use std::io::{BufWriter, Read, Write}; use std::io::{BufWriter, Read, Write};
use std::path::{Path, PathBuf}; use std::path::{Path, PathBuf};
+135 -154
View File
@@ -53,6 +53,8 @@ pub struct ImageData {
pub struct DocxHandler { pub struct DocxHandler {
temp_dir: PathBuf, temp_dir: PathBuf,
pub documents: std::collections::HashMap<String, DocxMetadata>, pub documents: std::collections::HashMap<String, DocxMetadata>,
// In-memory operations for documents created via this handler
in_memory_ops: std::collections::HashMap<String, Vec<DocxOp>>,
} }
impl DocxHandler { impl DocxHandler {
@@ -63,6 +65,7 @@ impl DocxHandler {
Ok(Self { Ok(Self {
temp_dir, temp_dir,
documents: std::collections::HashMap::new(), documents: std::collections::HashMap::new(),
in_memory_ops: std::collections::HashMap::new(),
}) })
} }
@@ -74,6 +77,7 @@ impl DocxHandler {
Ok(Self { Ok(Self {
temp_dir, temp_dir,
documents: std::collections::HashMap::new(), documents: std::collections::HashMap::new(),
in_memory_ops: std::collections::HashMap::new(),
}) })
} }
@@ -81,6 +85,7 @@ impl DocxHandler {
let doc_id = Uuid::new_v4().to_string(); let doc_id = Uuid::new_v4().to_string();
let doc_path = self.temp_dir.join(format!("{}.docx", doc_id)); let doc_path = self.temp_dir.join(format!("{}.docx", doc_id));
// Initialize empty document on disk
let docx = Docx::new(); let docx = Docx::new();
let file = File::create(&doc_path)?; let file = File::create(&doc_path)?;
docx.build().pack(file)?; docx.build().pack(file)?;
@@ -99,6 +104,7 @@ impl DocxHandler {
}; };
self.documents.insert(doc_id.clone(), metadata); self.documents.insert(doc_id.clone(), metadata);
self.in_memory_ops.insert(doc_id.clone(), Vec::new());
info!("Created new document with ID: {}", doc_id); info!("Created new document with ID: {}", doc_id);
Ok(doc_id) Ok(doc_id)
@@ -133,54 +139,10 @@ impl DocxHandler {
} }
pub fn add_paragraph(&mut self, doc_id: &str, text: &str, style: Option<DocxStyle>) -> Result<()> { pub fn add_paragraph(&mut self, doc_id: &str, text: &str, style: Option<DocxStyle>) -> Result<()> {
let metadata = self.documents.get(doc_id) self.ensure_modifiable(doc_id)?;
.ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?; let ops = self.in_memory_ops.get_mut(doc_id).unwrap();
ops.push(DocxOp::Paragraph { text: text.to_string(), style });
let mut file = File::open(&metadata.path)?; self.write_docx(doc_id)?;
let mut buffer = Vec::new();
file.read_to_end(&mut buffer)?;
let mut docx = Docx::from_reader(&buffer[..])?;
let mut paragraph = Paragraph::new().add_run(Run::new().add_text(text));
if let Some(style) = style {
let mut run = Run::new().add_text(text);
if let Some(size) = style.font_size {
run = run.size(size);
}
if style.bold == Some(true) {
run = run.bold();
}
if style.italic == Some(true) {
run = run.italic();
}
if style.underline == Some(true) {
run = run.underline("single");
}
if let Some(color) = style.color {
run = run.color(color);
}
paragraph = Paragraph::new().add_run(run);
if let Some(alignment) = style.alignment {
paragraph = match alignment.as_str() {
"left" => paragraph.align(AlignmentType::Left),
"center" => paragraph.align(AlignmentType::Center),
"right" => paragraph.align(AlignmentType::Right),
"justify" => paragraph.align(AlignmentType::Justified),
_ => paragraph,
};
}
}
docx = docx.add_paragraph(paragraph);
let file = File::create(&metadata.path)?;
docx.build().pack(file)?;
info!("Added paragraph to document {}", doc_id); info!("Added paragraph to document {}", doc_id);
Ok(()) Ok(())
} }
@@ -189,12 +151,6 @@ impl DocxHandler {
let metadata = self.documents.get(doc_id) let metadata = self.documents.get(doc_id)
.ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?; .ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?;
let mut file = File::open(&metadata.path)?;
let mut buffer = Vec::new();
file.read_to_end(&mut buffer)?;
let mut docx = Docx::from_reader(&buffer[..])?;
let heading_style = match level { let heading_style = match level {
1 => "Heading1", 1 => "Heading1",
2 => "Heading2", 2 => "Heading2",
@@ -204,16 +160,10 @@ impl DocxHandler {
6 => "Heading6", 6 => "Heading6",
_ => "Heading1", _ => "Heading1",
}; };
self.ensure_modifiable(doc_id)?;
let paragraph = Paragraph::new() let ops = self.in_memory_ops.get_mut(doc_id).unwrap();
.add_run(Run::new().add_text(text)) ops.push(DocxOp::Heading { text: text.to_string(), style: heading_style.to_string() });
.style(heading_style); self.write_docx(doc_id)?;
docx = docx.add_paragraph(paragraph);
let file = File::create(&metadata.path)?;
docx.build().pack(file)?;
info!("Added heading level {} to document {}", level, doc_id); info!("Added heading level {} to document {}", level, doc_id);
Ok(()) Ok(())
} }
@@ -222,35 +172,10 @@ impl DocxHandler {
let metadata = self.documents.get(doc_id) let metadata = self.documents.get(doc_id)
.ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?; .ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?;
let mut file = File::open(&metadata.path)?; self.ensure_modifiable(doc_id)?;
let mut buffer = Vec::new(); let ops = self.in_memory_ops.get_mut(doc_id).unwrap();
file.read_to_end(&mut buffer)?; ops.push(DocxOp::Table { data: table_data });
self.write_docx(doc_id)?;
let mut docx = Docx::from_reader(&buffer[..])?;
let col_count = table_data.rows.get(0).map(|r| r.len()).unwrap_or(0);
let mut table = Table::new(vec![TableCell::new(); col_count]);
for row_data in table_data.rows {
let mut cells = Vec::new();
for cell_text in row_data {
let cell = TableCell::new()
.add_paragraph(Paragraph::new().add_run(Run::new().add_text(cell_text)));
cells.push(cell);
}
while cells.len() < col_count {
cells.push(TableCell::new());
}
table = table.add_row(TableRow::new(cells));
}
docx = docx.add_table(table);
let file = File::create(&metadata.path)?;
docx.build().pack(file)?;
info!("Added table to document {}", doc_id); info!("Added table to document {}", doc_id);
Ok(()) Ok(())
} }
@@ -259,25 +184,10 @@ impl DocxHandler {
let metadata = self.documents.get(doc_id) let metadata = self.documents.get(doc_id)
.ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?; .ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?;
let mut file = File::open(&metadata.path)?; self.ensure_modifiable(doc_id)?;
let mut buffer = Vec::new(); let ops = self.in_memory_ops.get_mut(doc_id).unwrap();
file.read_to_end(&mut buffer)?; ops.push(DocxOp::List { items, ordered });
self.write_docx(doc_id)?;
let mut docx = Docx::from_reader(&buffer[..])?;
let numbering_id = if ordered { 1 } else { 2 };
for item in items {
let paragraph = Paragraph::new()
.add_run(Run::new().add_text(item))
.numbering(NumberingId::new(numbering_id), IndentLevel::new(0));
docx = docx.add_paragraph(paragraph);
}
let file = File::create(&metadata.path)?;
docx.build().pack(file)?;
info!("Added {} list to document {}", if ordered { "ordered" } else { "unordered" }, doc_id); info!("Added {} list to document {}", if ordered { "ordered" } else { "unordered" }, doc_id);
Ok(()) Ok(())
} }
@@ -286,18 +196,10 @@ impl DocxHandler {
let metadata = self.documents.get(doc_id) let metadata = self.documents.get(doc_id)
.ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?; .ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?;
let mut file = File::open(&metadata.path)?; self.ensure_modifiable(doc_id)?;
let mut buffer = Vec::new(); let ops = self.in_memory_ops.get_mut(doc_id).unwrap();
file.read_to_end(&mut buffer)?; ops.push(DocxOp::PageBreak);
self.write_docx(doc_id)?;
let mut docx = Docx::from_reader(&buffer[..])?;
let paragraph = Paragraph::new().add_run(Run::new().add_break(BreakType::Page));
docx = docx.add_paragraph(paragraph);
let file = File::create(&metadata.path)?;
docx.build().pack(file)?;
info!("Added page break to document {}", doc_id); info!("Added page break to document {}", doc_id);
Ok(()) Ok(())
} }
@@ -306,21 +208,10 @@ impl DocxHandler {
let metadata = self.documents.get(doc_id) let metadata = self.documents.get(doc_id)
.ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?; .ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?;
let mut file = File::open(&metadata.path)?; self.ensure_modifiable(doc_id)?;
let mut buffer = Vec::new(); let ops = self.in_memory_ops.get_mut(doc_id).unwrap();
file.read_to_end(&mut buffer)?; ops.push(DocxOp::Header(text.to_string()));
self.write_docx(doc_id)?;
let mut docx = Docx::from_reader(&buffer[..])?;
let header = Header::new().add_paragraph(
Paragraph::new().add_run(Run::new().add_text(text))
);
docx = docx.header(header);
let file = File::create(&metadata.path)?;
docx.build().pack(file)?;
info!("Set header for document {}", doc_id); info!("Set header for document {}", doc_id);
Ok(()) Ok(())
} }
@@ -329,21 +220,10 @@ impl DocxHandler {
let metadata = self.documents.get(doc_id) let metadata = self.documents.get(doc_id)
.ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?; .ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?;
let mut file = File::open(&metadata.path)?; self.ensure_modifiable(doc_id)?;
let mut buffer = Vec::new(); let ops = self.in_memory_ops.get_mut(doc_id).unwrap();
file.read_to_end(&mut buffer)?; ops.push(DocxOp::Footer(text.to_string()));
self.write_docx(doc_id)?;
let mut docx = Docx::from_reader(&buffer[..])?;
let footer = Footer::new().add_paragraph(
Paragraph::new().add_run(Run::new().add_text(text))
);
docx = docx.footer(footer);
let file = File::create(&metadata.path)?;
docx.build().pack(file)?;
info!("Set footer for document {}", doc_id); info!("Set footer for document {}", doc_id);
Ok(()) Ok(())
} }
@@ -397,6 +277,7 @@ impl DocxHandler {
if metadata.path.exists() { if metadata.path.exists() {
fs::remove_file(&metadata.path)?; fs::remove_file(&metadata.path)?;
} }
self.in_memory_ops.remove(doc_id);
info!("Closed document {}", doc_id); info!("Closed document {}", doc_id);
Ok(()) Ok(())
@@ -406,3 +287,103 @@ impl DocxHandler {
self.documents.values().cloned().collect() self.documents.values().cloned().collect()
} }
} }
#[derive(Debug, Clone)]
enum DocxOp {
Paragraph { text: String, style: Option<DocxStyle> },
Heading { text: String, style: String },
Table { data: TableData },
List { items: Vec<String>, ordered: bool },
PageBreak,
Header(String),
Footer(String),
}
impl DocxHandler {
fn ensure_modifiable(&self, doc_id: &str) -> Result<()> {
if !self.in_memory_ops.contains_key(doc_id) {
anyhow::bail!("Modifications are supported only for documents created by this server (doc_id: {})", doc_id);
}
Ok(())
}
fn write_docx(&self, doc_id: &str) -> Result<()> {
let metadata = self.documents.get(doc_id)
.ok_or_else(|| anyhow::anyhow!("Document not found: {}", doc_id))?;
let ops = self.in_memory_ops.get(doc_id)
.ok_or_else(|| anyhow::anyhow!("No in-memory ops for document: {}", doc_id))?;
let mut docx = Docx::new();
let mut header_text: Option<String> = None;
let mut footer_text: Option<String> = None;
for op in ops {
match op {
DocxOp::Paragraph { text, style } => {
let mut run = Run::new().add_text(text);
if let Some(st) = style {
if let Some(size) = st.font_size { run = run.size(size); }
if st.bold == Some(true) { run = run.bold(); }
if st.italic == Some(true) { run = run.italic(); }
if st.underline == Some(true) { run = run.underline("single"); }
if let Some(color) = &st.color { run = run.color(color.clone()); }
}
let para = Paragraph::new().add_run(run);
docx = docx.add_paragraph(para);
}
DocxOp::Heading { text, style } => {
let para = Paragraph::new().add_run(Run::new().add_text(text)).style(style);
docx = docx.add_paragraph(para);
}
DocxOp::Table { data } => {
let col_count = data.rows.get(0).map(|r| r.len()).unwrap_or(0);
// Build rows
let mut table = Table::new(vec![]);
for row in &data.rows {
let mut cells: Vec<TableCell> = Vec::new();
for cell_text in row {
let cell = TableCell::new().add_paragraph(Paragraph::new().add_run(Run::new().add_text(cell_text)));
cells.push(cell);
}
while cells.len() < col_count { cells.push(TableCell::new()); }
table = table.add_row(TableRow::new(cells));
}
docx = docx.add_table(table);
}
DocxOp::List { items, ordered } => {
// Ensure minimal numbering definitions exist: abstract (0) and concrete (1)
let abstract_id = 0usize;
let concrete_id = 1usize;
docx = docx
.add_abstract_numbering(docx_rs::AbstractNumbering::new(abstract_id))
.add_numbering(docx_rs::Numbering::new(concrete_id, abstract_id));
for item in items {
let para = Paragraph::new()
.add_run(Run::new().add_text(item))
.numbering(NumberingId::new(concrete_id), IndentLevel::new(0));
docx = docx.add_paragraph(para);
}
}
DocxOp::PageBreak => {
let para = Paragraph::new().add_run(Run::new().add_break(BreakType::Page));
docx = docx.add_paragraph(para);
}
DocxOp::Header(text) => { header_text = Some(text.clone()); }
DocxOp::Footer(text) => { footer_text = Some(text.clone()); }
}
}
if let Some(h) = header_text {
let header = Header::new().add_paragraph(Paragraph::new().add_run(Run::new().add_text(h)));
docx = docx.header(header);
}
if let Some(f) = footer_text {
let footer = Footer::new().add_paragraph(Paragraph::new().add_run(Run::new().add_text(f)));
docx = docx.footer(footer);
}
let file = File::create(&metadata.path)?;
docx.build().pack(file)?;
Ok(())
}
}
+38 -6
View File
@@ -11,12 +11,15 @@ use anyhow::Result;
use crate::docx_handler::{DocxHandler, DocxStyle, TableData, ImageData}; use crate::docx_handler::{DocxHandler, DocxStyle, TableData, ImageData};
use crate::converter::DocumentConverter; use crate::converter::DocumentConverter;
#[cfg(feature = "advanced-docx")]
use crate::advanced_docx::AdvancedDocxHandler; use crate::advanced_docx::AdvancedDocxHandler;
use crate::security::{SecurityConfig, SecurityMiddleware}; use crate::security::{SecurityConfig, SecurityMiddleware};
#[derive(Clone)]
pub struct DocxToolsProvider { pub struct DocxToolsProvider {
handler: Arc<Mutex<DocxHandler>>, handler: Arc<Mutex<DocxHandler>>,
converter: Arc<DocumentConverter>, converter: Arc<DocumentConverter>,
#[cfg(feature = "advanced-docx")]
advanced: Arc<AdvancedDocxHandler>, advanced: Arc<AdvancedDocxHandler>,
security: Arc<SecurityMiddleware>, security: Arc<SecurityMiddleware>,
security_config: SecurityConfig, security_config: SecurityConfig,
@@ -31,6 +34,7 @@ impl DocxToolsProvider {
Self { Self {
handler: Arc::new(Mutex::new(DocxHandler::new().expect("Failed to create DocxHandler"))), handler: Arc::new(Mutex::new(DocxHandler::new().expect("Failed to create DocxHandler"))),
converter: Arc::new(DocumentConverter::new()), converter: Arc::new(DocumentConverter::new()),
#[cfg(feature = "advanced-docx")]
advanced: Arc::new(AdvancedDocxHandler::new()), advanced: Arc::new(AdvancedDocxHandler::new()),
security: Arc::new(SecurityMiddleware::new(security_config.clone())), security: Arc::new(SecurityMiddleware::new(security_config.clone())),
security_config, security_config,
@@ -38,9 +42,8 @@ impl DocxToolsProvider {
} }
} }
#[async_trait] impl DocxToolsProvider {
impl ToolProvider for DocxToolsProvider { pub async fn list_tools(&self) -> Vec<Tool> {
async fn list_tools(&self) -> Vec<Tool> {
let mut all_tools = vec![ let mut all_tools = vec![
Tool { Tool {
name: "create_document".to_string(), name: "create_document".to_string(),
@@ -50,6 +53,7 @@ impl ToolProvider for DocxToolsProvider {
"properties": {}, "properties": {},
"required": [] "required": []
}), }),
annotations: None,
}, },
Tool { Tool {
name: "open_document".to_string(), name: "open_document".to_string(),
@@ -64,6 +68,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["path"] "required": ["path"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "add_paragraph".to_string(), name: "add_paragraph".to_string(),
@@ -98,6 +103,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "text"] "required": ["document_id", "text"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "add_heading".to_string(), name: "add_heading".to_string(),
@@ -122,6 +128,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "text", "level"] "required": ["document_id", "text", "level"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "add_table".to_string(), name: "add_table".to_string(),
@@ -153,6 +160,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "rows"] "required": ["document_id", "rows"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "add_list".to_string(), name: "add_list".to_string(),
@@ -177,6 +185,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "items"] "required": ["document_id", "items"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "add_page_break".to_string(), name: "add_page_break".to_string(),
@@ -191,6 +200,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id"] "required": ["document_id"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "set_header".to_string(), name: "set_header".to_string(),
@@ -209,6 +219,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "text"] "required": ["document_id", "text"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "set_footer".to_string(), name: "set_footer".to_string(),
@@ -227,6 +238,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "text"] "required": ["document_id", "text"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "find_and_replace".to_string(), name: "find_and_replace".to_string(),
@@ -249,6 +261,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "find_text", "replace_text"] "required": ["document_id", "find_text", "replace_text"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "extract_text".to_string(), name: "extract_text".to_string(),
@@ -263,6 +276,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id"] "required": ["document_id"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "get_metadata".to_string(), name: "get_metadata".to_string(),
@@ -277,6 +291,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id"] "required": ["document_id"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "save_document".to_string(), name: "save_document".to_string(),
@@ -295,6 +310,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "output_path"] "required": ["document_id", "output_path"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "close_document".to_string(), name: "close_document".to_string(),
@@ -309,6 +325,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id"] "required": ["document_id"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "list_documents".to_string(), name: "list_documents".to_string(),
@@ -318,6 +335,7 @@ impl ToolProvider for DocxToolsProvider {
"properties": {}, "properties": {},
"required": [] "required": []
}), }),
annotations: None,
}, },
Tool { Tool {
name: "convert_to_pdf".to_string(), name: "convert_to_pdf".to_string(),
@@ -336,6 +354,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "output_path"] "required": ["document_id", "output_path"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "convert_to_images".to_string(), name: "convert_to_images".to_string(),
@@ -367,7 +386,11 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "output_dir"] "required": ["document_id", "output_dir"]
}), }),
annotations: None,
}, },
// Advanced tools are gated and added only when feature is enabled
#[cfg(feature = "advanced-docx")]
Tool { Tool {
name: "merge_documents".to_string(), name: "merge_documents".to_string(),
description: Some("Merge multiple DOCX documents into one".to_string()), description: Some("Merge multiple DOCX documents into one".to_string()),
@@ -386,7 +409,9 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_ids", "output_path"] "required": ["document_ids", "output_path"]
}), }),
annotations: None,
}, },
#[cfg(feature = "advanced-docx")]
Tool { Tool {
name: "split_document".to_string(), name: "split_document".to_string(),
description: Some("Split a document at page breaks".to_string()), description: Some("Split a document at page breaks".to_string()),
@@ -404,6 +429,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "output_dir"] "required": ["document_id", "output_dir"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "get_document_structure".to_string(), name: "get_document_structure".to_string(),
@@ -418,6 +444,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id"] "required": ["document_id"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "analyze_formatting".to_string(), name: "analyze_formatting".to_string(),
@@ -432,6 +459,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id"] "required": ["document_id"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "get_word_count".to_string(), name: "get_word_count".to_string(),
@@ -446,6 +474,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id"] "required": ["document_id"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "search_text".to_string(), name: "search_text".to_string(),
@@ -474,6 +503,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "search_term"] "required": ["document_id", "search_term"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "export_to_markdown".to_string(), name: "export_to_markdown".to_string(),
@@ -492,6 +522,7 @@ impl ToolProvider for DocxToolsProvider {
}, },
"required": ["document_id", "output_path"] "required": ["document_id", "output_path"]
}), }),
annotations: None,
}, },
Tool { Tool {
name: "get_security_info".to_string(), name: "get_security_info".to_string(),
@@ -501,6 +532,7 @@ impl ToolProvider for DocxToolsProvider {
"properties": {}, "properties": {},
"required": [] "required": []
}), }),
annotations: None,
}, },
]; ];
@@ -513,7 +545,7 @@ impl ToolProvider for DocxToolsProvider {
all_tools all_tools
} }
async fn call_tool(&self, name: &str, arguments: Value) -> CallToolResponse { pub async fn call_tool(&self, name: &str, arguments: Value) -> CallToolResponse {
debug!("Calling tool: {} with arguments: {:?}", name, arguments); debug!("Calling tool: {} with arguments: {:?}", name, arguments);
// Security check // Security check
@@ -982,7 +1014,7 @@ impl ToolProvider for DocxToolsProvider {
let handler = self.handler.lock().unwrap(); let handler = self.handler.lock().unwrap();
match handler.extract_text(doc_id) { match handler.extract_text(doc_id) {
Ok(text) => { Ok(text) => {
let search_text = if case_sensitive { text } else { text.to_lowercase() }; let search_text = if case_sensitive { text.clone() } else { text.to_lowercase() };
let search_for = if case_sensitive { search_term.to_string() } else { search_term.to_lowercase() }; let search_for = if case_sensitive { search_term.to_string() } else { search_term.to_lowercase() };
let mut matches = Vec::new(); let mut matches = Vec::new();
@@ -1086,7 +1118,7 @@ impl ToolProvider for DocxToolsProvider {
_ => { _ => {
json!({ json!({
"success": false, "success": false,
"error": format!("Unknown tool: {}", name) "error": format!("Unknown or unsupported tool: {}", name)
}) })
} }
}; };
+56 -3
View File
@@ -13,7 +13,7 @@ mod docx_handler;
mod converter; mod converter;
#[cfg(feature = "runtime-server")] #[cfg(feature = "runtime-server")]
mod pure_converter; mod pure_converter;
#[cfg(feature = "runtime-server")] #[cfg(all(feature = "runtime-server", feature = "advanced-docx"))]
mod advanced_docx; mod advanced_docx;
mod security; mod security;
@@ -55,11 +55,64 @@ async fn main() -> Result<()> {
#[cfg(feature = "runtime-server")] #[cfg(feature = "runtime-server")]
{ {
use mcp_server::{Router, Server};
use mcp_server::router::RouterService;
use mcp_server::router::CapabilitiesBuilder;
use mcp_spec::{prompt::Prompt, resource::Resource};
use mcp_spec::protocol::ServerCapabilities;
use mcp_spec::content::Content;
use mcp_spec::tool::Tool as SpecTool;
use serde_json::Value as JsonValue;
use std::pin::Pin;
use std::future::Future;
use tokio::io::{stdin, stdout};
let security_config = security::SecurityConfig::from_args(args); let security_config = security::SecurityConfig::from_args(args);
info!("Starting DOCX MCP Server - Security: {}", security_config.get_summary()); info!("Starting DOCX MCP Server - Security: {}", security_config.get_summary());
// TODO: Integrate with mcp-server Router here. For now, just exit successfully. #[derive(Clone)]
info!("Server integration pending refactor; exiting."); struct DocxRouter(docx_tools::DocxToolsProvider);
impl Router for DocxRouter {
fn name(&self) -> String { "docx-mcp-server".to_string() }
fn instructions(&self) -> String { "DOCX tools for reading and exporting".to_string() }
fn capabilities(&self) -> ServerCapabilities {
CapabilitiesBuilder::new().with_tools(true).build()
}
fn list_tools(&self) -> Vec<SpecTool> {
// DocxToolsProvider::list_tools is async; block briefly with tokio runtime handle
let rt = tokio::runtime::Handle::current();
let tools = rt.block_on(self.0.list_tools());
tools.into_iter().map(|t| SpecTool{ name: t.name, description: t.description.unwrap_or_default(), input_schema: t.input_schema }).collect()
}
fn call_tool(&self, tool_name: &str, arguments: JsonValue) -> Pin<Box<dyn Future<Output = Result<Vec<Content>, mcp_spec::handler::ToolError>> + Send + 'static>> {
let provider = self.0.clone();
let name = tool_name.to_string();
Box::pin(async move {
let resp = provider.call_tool(&name, arguments).await;
// Convert our CallToolResponse (text JSON) to Content::text
let text = match resp.content.get(0) {
Some(mcp_core::types::ToolResponseContent::Text(t)) => t.text.clone(),
_ => serde_json::to_string(&resp).unwrap_or_else(|_| "{}".to_string()),
};
Ok(vec![Content::text(text)])
})
}
fn list_resources(&self) -> Vec<Resource> { vec![] }
fn read_resource(&self, _uri: &str) -> Pin<Box<dyn Future<Output = Result<String, mcp_spec::handler::ResourceError>> + Send + 'static>> {
Box::pin(async { Ok(String::new()) })
}
fn list_prompts(&self) -> Vec<Prompt> { vec![] }
fn get_prompt(&self, _prompt_name: &str) -> Pin<Box<dyn Future<Output = Result<String, mcp_spec::handler::PromptError>> + Send + 'static>> {
Box::pin(async { Ok(String::new()) })
}
}
let router = DocxRouter(DocxToolsProvider::new_with_security(security_config));
let service = RouterService(router);
let server = Server::new(service);
let transport = mcp_server::ByteTransport::new(stdin(), stdout());
server.run(transport).await?;
} }
#[cfg(not(feature = "runtime-server"))] #[cfg(not(feature = "runtime-server"))]
+7 -7
View File
@@ -9,7 +9,7 @@ use tracing::{debug, info, warn};
use roxmltree; use roxmltree;
use zip::ZipArchive; use zip::ZipArchive;
use rusttype::{Font, Scale}; use rusttype::{Font, Scale};
use lopdf::{self, dictionary, Object}; use ::lopdf::{self as lopdf_crate, dictionary, Object};
pub struct PureRustConverter; pub struct PureRustConverter;
@@ -244,7 +244,7 @@ impl PureRustConverter {
/// Merge multiple PDFs using pure Rust /// Merge multiple PDFs using pure Rust
pub fn merge_pdfs_pure(&self, pdf_paths: &[PathBuf], output_path: &Path) -> Result<()> { pub fn merge_pdfs_pure(&self, pdf_paths: &[PathBuf], output_path: &Path) -> Result<()> {
use lopdf::{Document, Object, ObjectId}; use ::lopdf::{Document, Object, ObjectId};
// Create a new document for merging // Create a new document for merging
let mut merged_doc = Document::with_version("1.5"); let mut merged_doc = Document::with_version("1.5");
@@ -270,7 +270,7 @@ impl PureRustConverter {
// Build the page tree for merged document // Build the page tree for merged document
let pages_id = merged_doc.new_object_id(); let pages_id = merged_doc.new_object_id();
let pages_dict = lopdf::dictionary! { let pages_dict = ::lopdf::dictionary! {
"Type" => "Pages", "Type" => "Pages",
"Kids" => all_pages.iter().map(|id| Object::Reference(*id)).collect::<Vec<_>>(), "Kids" => all_pages.iter().map(|id| Object::Reference(*id)).collect::<Vec<_>>(),
"Count" => all_pages.len() as i32, "Count" => all_pages.len() as i32,
@@ -279,7 +279,7 @@ impl PureRustConverter {
// Update catalog // Update catalog
let catalog_id = merged_doc.new_object_id(); let catalog_id = merged_doc.new_object_id();
let catalog = lopdf::dictionary! { let catalog = ::lopdf::dictionary! {
"Type" => "Catalog", "Type" => "Catalog",
"Pages" => Object::Reference(pages_id), "Pages" => Object::Reference(pages_id),
}; };
@@ -295,7 +295,7 @@ impl PureRustConverter {
/// Split a PDF into individual pages using pure Rust /// Split a PDF into individual pages using pure Rust
pub fn split_pdf_pure(&self, pdf_path: &Path, output_dir: &Path) -> Result<Vec<PathBuf>> { pub fn split_pdf_pure(&self, pdf_path: &Path, output_dir: &Path) -> Result<Vec<PathBuf>> {
use lopdf::Document; use ::lopdf::Document;
fs::create_dir_all(output_dir)?; fs::create_dir_all(output_dir)?;
@@ -314,7 +314,7 @@ impl PureRustConverter {
// Create page tree // Create page tree
let pages_id = single_page_doc.new_object_id(); let pages_id = single_page_doc.new_object_id();
let pages_dict = lopdf::dictionary! { let pages_dict = ::lopdf::dictionary! {
"Type" => "Pages", "Type" => "Pages",
"Kids" => vec![Object::Reference(new_page_id)], "Kids" => vec![Object::Reference(new_page_id)],
"Count" => 1, "Count" => 1,
@@ -323,7 +323,7 @@ impl PureRustConverter {
// Create catalog // Create catalog
let catalog_id = single_page_doc.new_object_id(); let catalog_id = single_page_doc.new_object_id();
let catalog = lopdf::dictionary! { let catalog = ::lopdf::dictionary! {
"Type" => "Catalog", "Type" => "Catalog",
"Pages" => Object::Reference(pages_id), "Pages" => Object::Reference(pages_id),
}; };