commit dd02658: [Minor] Pdf: Improve debug logging

[Vsevolod Stakhov]

Author: Vsevolod Stakhov
Date: 2021-01-11 19:46:38 +0000
URL: https://github.com/rspamd/rspamd/commit/dd026588fc294a86bea3922e65836dbe1a7cacd6 (HEAD -> master)

[Minor] Pdf: Improve debug logging

---
 lualib/lua_content/pdf.lua | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/lualib/lua_content/pdf.lua b/lualib/lua_content/pdf.lua
index a5e23d11f..b3c1f6002 100644
--- a/lualib/lua_content/pdf.lua
+++ b/lualib/lua_content/pdf.lua
@@ -519,7 +519,7 @@ end
 local process_dict
 
 -- This function processes javascript string and returns JS hash and JS rspamd_text
-local function process_javascript(task, pdf, js)
+local function process_javascript(task, pdf, js, obj)
   local rspamd_cryptobox_hash = require "rspamd_cryptobox_hash"
   if type(js) == 'string' then
     js = rspamd_text.fromstring(js):oneline()
@@ -545,6 +545,7 @@ local function process_javascript(task, pdf, js)
     data = js,
     hash = hash:hex(),
     bin_hash = bin_hash,
+    object = obj,
   }
   pdf.scripts[bin_hash] = njs
   return njs
@@ -567,7 +568,7 @@ local function process_action(task, pdf, obj)
         end
       end
 
-      js = process_javascript(task, pdf, js)
+      js = process_javascript(task, pdf, js, obj)
       if js then
         obj.js = js
         lua_util.debugm(N, task, 'extracted javascript from %s:%s: %s',
@@ -614,6 +615,7 @@ local function process_catalog(task, pdf, obj)
           lua_util.debugm(N, task, 'found openaction JS in %s:%s: %s',
               obj.major, obj.minor, action.js)
           pdf.openaction = action.js
+          action.js.object = obj
         elseif action.launch then
           lua_util.debugm(N, task, 'found openaction launch in %s:%s: %s',
               obj.major, obj.minor, action.launch)
@@ -765,7 +767,7 @@ process_dict = function(task, pdf, obj, dict)
           end
         end
 
-        js = process_javascript(task, pdf, js)
+        js = process_javascript(task, pdf, js, obj)
         if js then
           obj.js = js
           lua_util.debugm(N, task, 'extracted javascript from %s:%s: %s',
@@ -1243,8 +1245,10 @@ local function process_pdf(input, mpart, task)
           -- OpenAction only
           if pdf_object.openaction and pdf_object.openaction.bin_hash then
             if config.min_js_fuzzy and #pdf_object.openaction.data >= config.min_js_fuzzy then
-              lua_util.debugm(N, task, "pdf: add fuzzy hash from openaction: %s",
-                  pdf_object.openaction.hash)
+              lua_util.debugm(N, task, "pdf: add fuzzy hash from openaction: %s; size = %s; object: %s:%s",
+                  pdf_object.openaction.hash,
+                  #pdf_object.openaction.data,
+                  pdf_object.openaction.object.major, pdf_object.openaction.object.minor)
               table.insert(pdf_output.fuzzy_hashes, pdf_object.openaction.bin_hash)
             else
               lua_util.debugm(N, task, "pdf: skip fuzzy hash from Javascript: %s, too short: %s",
@@ -1255,8 +1259,10 @@ local function process_pdf(input, mpart, task)
           -- All hashes
           for h,sc in pairs(pdf_object.scripts) do
             if config.min_js_fuzzy and #sc.data >= config.min_js_fuzzy then
-              lua_util.debugm(N, task, "pdf: add fuzzy hash from Javascript: %s",
-                  sc.hash)
+              lua_util.debugm(N, task, "pdf: add fuzzy hash from Javascript: %s; size = %s; object: %s:%s",
+                  sc.hash,
+                  #sc.data,
+                  sc.object.major, sc.object.minor)
               table.insert(pdf_output.fuzzy_hashes, h)
             else
               lua_util.debugm(N, task, "pdf: skip fuzzy hash from Javascript: %s, too short: %s",