Phrase Queries

"the quick brown fox" should match documents where these words appear IN ORDER, ADJACENT.

To support this, the inverted index stores positions along with doc IDs:

quick -> {doc1: [3, 17, 89], doc2: [4]}       # term appears at these positions
brown -> {doc1: [4, 18], doc2: [5]}
fox   -> {doc1: [5, 19], doc2: [6]}

For phrase "quick brown fox":

1. Look up each term's positions per doc.
2. Intersect docs (must contain all 3 terms).
3. For each remaining doc, find positions where:
   • quick at position p
   • brown at position p+1
   • fox at position p+2

Algorithm: for each pair of consecutive terms, check if any quick position p has brown at p+1, etc.

Variants: proximity search ("quick fox"~3 = words within 3 of each other), ordered proximity (must appear in this order, but with up to N words between).