From c6f41ce391b0f8f570dd8a46b253ab7b25a07b77 Mon Sep 17 00:00:00 2001
From: Julius Volz <julius.volz@gmail.com>
Date: Tue, 18 Jul 2017 12:35:41 +0200
Subject: [PATCH] Allow metric and label names to begin with underscores
 (#2961)

While this is not recommended, it is allowed in our data model
(https://prometheus.io/docs/concepts/data_model/).

Fixes https://github.com/prometheus/prometheus/issues/2959
---
 pkg/textparse/lex.l         | 13 ++++++-------
 pkg/textparse/lex.l.go      | 11 +++++------
 pkg/textparse/parse_test.go | 12 +++++++++++-
 3 files changed, 22 insertions(+), 14 deletions(-)

diff --git a/pkg/textparse/lex.l b/pkg/textparse/lex.l
index c79094dca..5fd8c6e37 100644
--- a/pkg/textparse/lex.l
+++ b/pkg/textparse/lex.l
@@ -39,7 +39,7 @@ const (
 // the end of the file.
 func (l *lexer) Lex() int {
     l.state = lstateInit
-    
+
     if l.i >= len(l.b) {
         return eof
     }
@@ -51,7 +51,6 @@ func (l *lexer) Lex() int {
 %}
 
 D     [0-9]
-S     [a-zA-Z]
 L     [a-zA-Z_]
 M     [a-zA-Z_:]
 
@@ -69,7 +68,7 @@ M     [a-zA-Z_:]
 #[^\r\n]*\n                           l.mstart = l.i
 [\r\n \t]+                            l.mstart = l.i
 
-{S}({M}|{D})*                         l.state = lstateName
+{M}({M}|{D})*                         l.state = lstateName
                                       l.offsets = append(l.offsets, l.i)
                                       l.mend = l.i
 
@@ -85,7 +84,7 @@ M     [a-zA-Z_:]
 <lstateLabels>(,?[ \t]*)              l.state = lstateLName
                                       l.offsets = append(l.offsets, l.i)
 
-<lstateLName>{S}({L}|{D})*            l.offsets = append(l.offsets, l.i)
+<lstateLName>{L}({L}|{D})*            l.offsets = append(l.offsets, l.i)
 <lstateLName>[ \t]*=                  l.state = lstateLValue
 
 <lstateLValue>[ \t]+
@@ -110,7 +109,7 @@ M     [a-zA-Z_:]
                                       }
                                       l.state = lstateTimestamp
 
-<lstateTimestamp>[ \t]+               l.tstart = l.i         
+<lstateTimestamp>[ \t]+               l.tstart = l.i
 <lstateTimestamp>{D}+                 ts, err := strconv.ParseInt(yoloString(l.b[l.tstart:l.i]), 10, 64)
                                       if err != nil {
                                           l.err = err
@@ -119,8 +118,8 @@ M     [a-zA-Z_:]
                                       l.ts = &ts
 <lstateTimestamp>[\r\n]+              l.nextMstart = l.i
                                       return 1
-<lstateTimestamp>\0                   return 1              
-   
+<lstateTimestamp>\0                   return 1
+
 %%
     l.err = fmt.Errorf("no token found")
     return -1
diff --git a/pkg/textparse/lex.l.go b/pkg/textparse/lex.l.go
index 00f204faa..1622a6421 100644
--- a/pkg/textparse/lex.l.go
+++ b/pkg/textparse/lex.l.go
@@ -83,12 +83,12 @@ yystart1:
 		goto yyabort
 	case c == '#':
 		goto yystate4
+	case c == ':' || c >= 'A' && c <= 'Z' || c == '_' || c >= 'a' && c <= 'z':
+		goto yystate6
 	case c == '\t' || c == '\n' || c == '\r' || c == ' ':
 		goto yystate3
 	case c == '\x00':
 		goto yystate2
-	case c >= 'A' && c <= 'Z' || c >= 'a' && c <= 'z':
-		goto yystate6
 	}
 
 yystate2:
@@ -327,7 +327,7 @@ yystart26:
 		goto yystate28
 	case c == '\t' || c == ' ':
 		goto yystate27
-	case c >= 'A' && c <= 'Z' || c >= 'a' && c <= 'z':
+	case c >= 'A' && c <= 'Z' || c == '_' || c >= 'a' && c <= 'z':
 		goto yystate29
 	}
 
@@ -436,7 +436,7 @@ yyrule3: // [\r\n \t]+
 		l.mstart = l.i
 		goto yystate0
 	}
-yyrule4: // {S}({M}|{D})*
+yyrule4: // {M}({M}|{D})*
 	{
 		l.state = lstateName
 		l.offsets = append(l.offsets, l.i)
@@ -469,7 +469,7 @@ yyrule9: // (,?[ \t]*)
 		l.offsets = append(l.offsets, l.i)
 		goto yystate0
 	}
-yyrule10: // {S}({L}|{D})*
+yyrule10: // {L}({L}|{D})*
 	{
 		l.offsets = append(l.offsets, l.i)
 		goto yystate0
@@ -543,7 +543,6 @@ yyrule20: // [\r\n]+
 yyrule21: // \0
 	{
 		return 1
-
 	}
 	panic("unreachable")
 
diff --git a/pkg/textparse/parse_test.go b/pkg/textparse/parse_test.go
index b1ef9cca0..25e11ef1f 100644
--- a/pkg/textparse/parse_test.go
+++ b/pkg/textparse/parse_test.go
@@ -43,7 +43,9 @@ go_gc_duration_seconds_count 99
 some:aggregate:rate5m{a_b="c"}	1
 # HELP go_goroutines Number of goroutines that currently exist.
 # TYPE go_goroutines gauge
-go_goroutines 33  	123123`
+go_goroutines 33  	123123
+_metric_starting_with_underscore 1
+testmetric{_label_starting_with_underscore="foo"} 1`
 	input += "\nnull_byte_metric{a=\"abc\x00\"} 1"
 
 	int64p := func(x int64) *int64 { return &x }
@@ -103,6 +105,14 @@ go_goroutines 33  	123123`
 			v:    33,
 			t:    int64p(123123),
 			lset: labels.FromStrings("__name__", "go_goroutines"),
+		}, {
+			m:    "_metric_starting_with_underscore",
+			v:    1,
+			lset: labels.FromStrings("__name__", "_metric_starting_with_underscore"),
+		}, {
+			m:    "testmetric{_label_starting_with_underscore=\"foo\"}",
+			v:    1,
+			lset: labels.FromStrings("__name__", "testmetric", "_label_starting_with_underscore", "foo"),
 		}, {
 			m:    "null_byte_metric{a=\"abc\x00\"}",
 			v:    1,