This study introduces a rule-based Praat script designed to generate P-ToBI labels based on the pitch contour given a tier with by-syllable intervals and stress marks. The system was trained on a 96-sentence corpus comprising all Nuclear Pitch Accents (NPA) and Boundary Tones (BT) in European Portuguese (EP). Evaluation was conducted on a separate corpus of 146 sentences showing a success rate of 73.8% (k=0.6) for NPA and 78.7% for BT (k=0.6). The qualitative analysis of errors, excluding those stemming from the pitch tracking algorithm, exposes challenges in accurately identifying falling NPAs, particularly instances of L*, H*+L, and H+L* followed by a low BT (although they can be accurately distinguished using an additive model). The performance of the system contrasts with results obtained with similar procedures for other Romance languages that get to 90% of success. We argue that the performance difference stems from principles underlying different ToBI systems (with P-ToBI being more phonological), and specificities of the phonological system of EP, namely word-final vowel reduction and deletion. This suggests that a rule-based approach relying solely on the acoustic signal may not be the most suitable for European Portuguese.