init

.gitignore

@@ -0,0 +1,2 @@
+.env
+data/

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "make-llm-files"]
+	path = make-llm-files
+	url = https://git.fredericofalcao.com/git/make-llm-files

make-llm-files

@@ -0,0 +1 @@
+Subproject commit 5a24d4946d3c53105b191bd9c43c5754bc71d0f1

src/Makefile

@@ -0,0 +1,72 @@
+include ../.env
+export
+
+.SECONDARY:
+
+
+SHELL=/bin/bash -x
+ROOT_DIR=/mnt/servers/aws.fredericofalcao.com/api.fredericofalcao.com/data/emails
+#SRC_EML_FILES=$(notdir $(shell find "$(ROOT_DIR)" -type f -name "*.eml" | head -n 2))
+LATEST_FILE:=$(shell find "$(ROOT_DIR)" -name "*.eml" -printf '%T@ %p\n' | sort -n | tail -1)
+LATEST_FILE_MODIFIED_TIME=$(shell echo "$(LATEST_FILE)" | cut -d' ' -f1)
+LATEST_FILE_FILE_PATH=$(shell echo "$(LATEST_FILE)" | cut -d' ' -f2-)
+
+SRC_EML_FILES=../data/$(notdir $(LATEST_FILE_FILE_PATH))
+SRC_TXT_FILES=$(subst .eml,.email.txt,$(SRC_EML_FILES))
+SRC_SCORE_FILES=$(subst .email.txt,.score.txt,$(SRC_TXT_FILES))
+SRC_FILES=$(subst .score.txt,.notification.txt,$(SRC_SCORE_FILES))
+	
+
+all: $(SRC_FILES)
+
+# 6. Run the MAIL-ROUTE script to either notify the user of a new message, or ignore
+%.notification.txt:  %.score.txt  %.header.txt %.summary.txt
+	TELEGRAM_BOT_TOKEN=$(TELEGRAM_BOT_TOKEN) TELEGRAM_CHAT_ID=$(TELEGRAM_CHAT_ID) ./mail-route $^ > $@
+
+%.header.txt: %.eml
+	echo > $@
+	cat $< | grep -i -e '^From:' | grep -o -E -e '@[a-zA-Z0-9_.-]+' | tr -d '@' | tr -d '\n' >> $@
+	printf " -> "  >> $@
+	cat $< | grep -i -e '^To:' | grep -o -E -e '[a-zA-Z0-9_.-]+@' | tr -d '@' | tr -d '\n' >> $@
+	printf "\n"  >> $@
+
+
+# 4. Generate a summary of the email
+%.summary.txt: %.email.txt
+	cd ../make-llm-files && make clean
+	echo "You are a machine executive assistant tasked with parsing emails. User will send the content of a received email, and you (agent) must send a one phrase (max 12 words) sentence as an executive summary of the email." > ../make-llm-files/01_root.sys
+	cp ../data/$< ../make-llm-files/user_message.txt
+	echo '{}' > ../make-llm-files/settings.json
+	cd ../make-llm-files && make 
+	mv ../make-llm-files/agent_message.txt $@
+
+
+
+# 3. Convert a text-only email to a score (something a script/email-router can react to)
+%.score.txt: %.email.txt
+	cd ../make-llm-files && make clean
+	echo "You are a machine executive assistant tasked with parsing emails urgency/importance. User will send the content of a received email, and you (agent) must REPLY ONLY with a single score [0.00-1.00] where you assess how important/urgent the email is." > ../make-llm-files/01_root.sys
+	cp ../data/$< ../make-llm-files/user_message.txt
+	echo '{"temperature": 0.8,"max_output_tokens":4}' > ../make-llm-files/settings.json
+	cd ../make-llm-files && make 
+	mv ../make-llm-files/agent_message.txt $@
+	
+# 2. Convert an eml to text-only (something that an llm can parse)
+%.email.txt: %.eml
+	./extract_text_from_eml.sh $< > $@
+
+# 1. Copy the email file (eml) locallly
+%.eml:
+	cp $(shell find $(ROOT_DIR) -name "$(notdir $@)") ../data/
+
+
+telegram_notification.log: email_message.eml
+	curl -s -X POST "https://api.telegram.org/bot$(TELEGRAM_BOT_TOKEN)/sendDocument" \
+  		-F chat_id="$(TELEGRAM_CHAT_ID)" \
+  		-F caption="Here is your file 📎" \
+  		-F document="@$<" > $@
+
+
+
+clean:
+	rm *.sys *.txt *.log *.eml

src/extract_text_from_eml.sh

@@ -0,0 +1,48 @@
+#!/bin/bash
+EMAIL_FILENAME=$1
+
+# REQUIRES: 
+#     sudo apt-get -y install w3m
+
+
+find_text() {
+
+# Non-multipart fallback (single-part message)
+if [[ "$ATTACHMENTS" == "Did not find anything to unpack from standard input" ]]; then
+    grep -A999999 -i -e '^From:' $EMAIL_FILENAME
+    return
+fi
+
+# Check if there is a TEXT/PLAIN part
+if [[ ! -z $(echo "$ATTACHMENTS" | grep "text/plain") ]]
+then 
+	# Check if it is empty
+	TEXT_PLAIN_FILE=$(echo "$ATTACHMENTS" | grep "text/plain" | awk '{print $1}')
+	if [[ ! -z "$(cat $T_DIR/$TEXT_PLAIN_FILE)" ]]
+	then
+		grep -i -e 'From:' $EMAIL_FILENAME
+		grep -i -e 'To:' $EMAIL_FILENAME
+		grep -i -e 'Subject:' $EMAIL_FILENAME
+		qprint -d $T_DIR/$TEXT_PLAIN_FILE
+		return
+	fi
+
+fi
+if [[ ! -z $(echo "$ATTACHMENTS" | grep "text/html") ]]
+then 
+	HTML_FILE=$(echo "$ATTACHMENTS" | grep "text/html" | awk '{print $1}')
+	# CONVERT HTML TO TEXT
+	qprint -d $T_DIR/$HTML_FILE | w3m -dump -I UTF-8 -O UTF-8 -T text/html -
+	return
+fi
+}
+
+
+T_DIR=$(mktemp -d)
+ATTACHMENTS=$(munpack -q -t -C $T_DIR < $EMAIL_FILENAME)
+
+find_text
+
+rm -rf $T_DIR;
+
+

src/mail-route

@@ -0,0 +1,82 @@
+#!/usr/bin/env php
+<?php
+
+
+/*
+## FREDS EMAIL SPAM SYSTEM
+
+# < 0.20 : delete / reject
+# < 0.40 : never notify / archive
+# < 0.60 : notify at end-of-week
+# < 0.80 : notify at end-of-day
+# < 1.00 : notify immediately
+
+*/
+
+$bin_name = array_shift($argv);
+$score = floatval(trim(file_get_contents(array_shift($argv))));
+$message = implode("\n",array_map(fn($o)=>trim(file_get_contents($o)),$argv)); 
+$telegram_bot_token = getenv()["TELEGRAM_BOT_TOKEN"] ?? ""; 
+$telegram_chat_id   = getenv()["TELEGRAM_CHAT_ID"] ?? "";
+
+function sendTelegramMessage($message, $telegram_bot_token, $telegram_chat_id) {
+     
+     // Build URL
+     $url = "https://api.telegram.org/bot{$telegram_bot_token}/sendMessage";
+     
+     // cURL request
+     $ch = curl_init($url);
+     curl_setopt_array($ch, [
+         CURLOPT_POST           => true,
+         CURLOPT_RETURNTRANSFER => true,
+         CURLOPT_HTTPHEADER     => ['Content-Type: application/json'],
+         CURLOPT_POSTFIELDS     => json_encode([
+             'chat_id' => $telegram_chat_id,
+             'text'    => $message,
+         ]),
+     ]);
+     
+     $response = curl_exec($ch);
+     
+     if ($response === false) {
+         return "cURL error: " . curl_error($ch) . PHP_EOL;
+     } else {
+         return $response ;
+     }
+     curl_close($ch);
+}
+
+if ( $score < 0.2 ) {
+
+	// DELETE / REJECT
+	sendTelegramMessage("_message rejected_", $telegram_bot_token, $telegram_chat_id);
+	sendTelegramMessage($message, $telegram_bot_token, $telegram_chat_id);
+
+
+} elseif ($score < 0.4) {
+
+	// NEVER NOTIFY
+	sendTelegramMessage("_message would be archived_", $telegram_bot_token, $telegram_chat_id);
+	sendTelegramMessage($message, $telegram_bot_token, $telegram_chat_id);
+
+} elseif ($score < 0.6) {
+
+	// NOTIFY AT THE END-OF-WEEK
+	sendTelegramMessage("_message is low-priority_", $telegram_bot_token, $telegram_chat_id);
+	sendTelegramMessage($message, $telegram_bot_token, $telegram_chat_id);
+
+
+} elseif ($score < 0.8) {
+
+	// NOTIFY AT THE END-OF-DAY
+	sendTelegramMessage("_message is medium-priority_", $telegram_bot_token, $telegram_chat_id);
+	sendTelegramMessage($message, $telegram_bot_token, $telegram_chat_id);
+
+} else {
+
+	// NOTIFY THE USER IMMEDIATELY
+	sendTelegramMessage("_message is high-priority_", $telegram_bot_token, $telegram_chat_id);
+	sendTelegramMessage($message, $telegram_bot_token, $telegram_chat_id);
+}
+
+